JP2005514098A - Method and system for assessing cardiac ischemia using an exercise protocol - Google Patents

Abstract

  Described herein are methods (40a-48a and 40b-48b) that provide a measure of a subject's heart or cardiovascular health and assess cardiac ischemia in the subject.

Description

[Related applications]
This application is a continuation-in-part of U.S. patent application Ser. No. 10 / 036,005, filed Dec. 26, 2001, which is also filed on Jun. 26, 2001. This is a continuation-in-part of pending US application Ser. No. 09 / 891,910, and this patent application is also a continuation of part of what was filed as US patent application Ser. No. 09 / 603,286 on June 26, 2000. It is an application. The entire disclosure of all these patents is hereby incorporated by reference.

[Field of the Invention]
The present invention relates to a non-invasive high-precision diagnosis of cardiac ischemia based on the processing of electrocardiogram (ECG) data on the body surface. The quantitative assessment method of the present invention for cardiac ischemia can generally indicate the health status of both the heart itself and the cardiovascular system at the same time.

  Heart attacks and other heart ischemia symptoms are the leading cause of death and disability in the United States. In general, the susceptibility of a particular patient to a heart attack, etc., is due to cardiac ischemia (insufficient blood flow to the heart tissue itself during periods of increased cardiac activity, resulting in inadequate oxygen supply) Can be evaluated by examining the evidence of Of course, it is highly desirable that the measurement technique be easy enough and can be performed without undue stress on the heart (its condition is not yet known) and without undue discomfort to the patient. It is also desirable that this measurement technique be useful for patients with various degrees of health, such as patients without detectable ischemia, patients with moderate ischemia, and patients with coronary artery disease.

  The cardiovascular system responds to changes in physiological stress by adjusting the heart rate. This adjustment can be evaluated by measuring the RR interval of the ECG on the body surface. The time interval between successive R waves indicates the interval between successive heartbeats (RR interval). This adjustment usually occurs with a corresponding change in the duration of the ECG QT interval. This QT interval characterizes the duration of myocardial electrical excitation and shows the duration of action potential averaged over a constant volume of myocardium (FIG. 1). Generally speaking, the duration of the average action potential measured as a QT interval at each ECG lead can be considered as an index of time-varying cardiac contraction activity.

  Recent advances in computer technology have led to improvements in automated analysis of heart rate and QT interval variability. Observing QT interval variability (variance) performed separately or in combination with heart rate (or RR interval) variability analysis provides an effective tool for assessing individual susceptibility to cardiac arrhythmias. This is now well known (B. Surawicz, J. Cardiovasc. Electrophysiol, 1996, 7, 777-784). Patent applications for another type of QT and other interval variations for cardiac arrhythmia susceptibility are Chamoun US Pat. No. 5,020,540, 1991; Wang US Pat. No. 4,870,974, 1989; Kroll et al. US Pat. No. 5,117,834, 1992; Henkin et al. US Pat. No. 5,323,783, 1994; Xue et al. US Pat. No. 5,792,065, 1998 Lander et al., US Pat. No. 5,827,195, 1998; Lander et al. US Pat. No. 5,891,047, 1999; Hojum et al. US Pat. No. 5,951,484, 1999; Explained.

  It has recently been discovered that cardiac electrical instability can also be predicted by linking observations of QT dispersion to analysis of ECG T-wave iterations (Verrier et al., US Pat. No. 5,560,370, 5,842,997, 5,921,940). This scheme is useful to some extent in identifying and managing individuals against the risk of sudden cardiac death. The inventors report that the dispersion of QT intervals is linked to the risk for arrhythmia in patients with long QT syndrome. However, it is said that it is difficult to accurately predict the electrical instability of the heart with only the dispersion of the QT interval without simultaneously measuring T wave repetitions (see US Pat. No. 5,560,370). 6, lines 4-15).

  Another application to analyze the variance of the QT interval for the prediction of sudden cardiac death is filed by J. Sarma (US Pat. No. 5,419,338). He explains how to test the autonomic nervous system. This method is designed to assess the imbalance between the control of both parasympathetic and sympathetic nerves to the heart, thereby indicating a tendency to sudden cardiac death.

  The same inventors have proposed that autonomic nervous system test procedures can be designed based on QT hysteresis (J. Sarma et al., PACE 10, 485-491 (1988)). Hysteresis between exercise and recovery was observed and was thought to be the result of sympathetic adrenal activity during the period after the initial exercise. Such activity appeared in the process of the QT interval adapting to changes in the RR interval during exercise with rapidly changing loads.

  The effect of sympathetic adrenal activity in the time course of rapid adaptation of the QT interval to rapid changes in RR interval dynamics and the apparent dependence of this hysteresis, suggests that myocardial electrical parameters and heart electrical conduction The sensitivity of this method to ischemic-like changes is fundamentally faded. For this reason, this type of hysteresis phenomenon is not useful for evaluating the health condition of the myocardium itself, that is, for evaluating cardiac ischemia.

  A similar sympathetic adrenal hysteretic phenomenon has been observed by A. Krahn et al. (Circulation 96, 1551-1556 (1997) (see FIG. 2)). The authors state that this type of QT interval hysteresis can be a marker for long QT syndrome. However, long QT syndrome hysteresis reflects a genetic defect in the ion channel in the heart associated with exercise or stress-induced syncope or sudden death. Thus, as in the previous example, for two different reasons, it is also not a measure of cardiac ischemia or myocardial ischemic health.

  A conventional non-invasive method for assessing coronary artery disease associated with cardiac ischemia is by observing morphological changes in the body surface electrocardiogram during physiological exercise (stress testing). Changes in ECG morphology, such as T-wave reversal, are well known for qualitative signs of ischemia. In addition to the rise or fall of the ST portion measured in relation to the average baseline, the dynamics of the ST portion of the ECG is continuously monitored while the shape and slope are changing in response to exercise load. . Comparison of some of these changes with the average value of the monitored ST segment data provides inadequate coronary blood circulation and an indication of the occurrence of ischemia. Despite the availability of computerized Holter-monitor devices that are widely clinically accepted and automatically process ST segment data, the diagnostic value of this method is low in sensitivity and resolution. Limited for. This approach is particularly reliable for ischemic events primarily associated with relatively large coronary artery occlusion, and its wide use often results in false positives. This in turn can lead to unnecessary and more expensive invasive cardiac catheterization.

  The relatively low sensitivity and low resolution, which are the fundamental disadvantages of the conventional ST partial descent method, are essentially based on measuring the amplitude of the ECG signal on the body surface. It is. This signal itself does not accurately reflect changes in the electrical parameters of individual heart cells that normally change during cardiac ischemic events. The ECG signal on the body surface is a complex determined by action potentials resulting from the discharge of hundreds of thousands of individual excited heart cells. If the electrical activity of the excited cells changes slightly locally during the progression of local ischemia resulting from exercise, the electrical image in the ECG signal on the body surface is from the rest of the heart. It is significantly diminished by the collective signal. Thus, regardless of physiological conditions such as stress or exercise, conventional body surface ECG data processing has a relatively high threshold of detectable ischemic morphological changes in the ECG signal (sensitivity). Is low). Accurate and complete identification of such changes is still an open signal processing problem.

  Thus, it is non-invasive, not overly uncomfortable or stressful, and can be performed with relatively simple equipment for patients of various health conditions, and can detect low levels of ischemia, There is a need to provide a technique for detecting and measuring cardiac ischemia in a patient.

  The present invention applies to all individual patients (not just those suffering from coronary artery disease) when the patient's heart rate reaches a sufficiently high level (this level varies from individual patient) It is based on the finding that slowing or stopping exercise (or other heart rate increasing load) slowly or rapidly does not elicit a clear or significant sympathetic adrenal response in the subject. As a result, exercise can be considered quasi-stationary, data can be collected from such patients, and the patient's cardiac ischemia and / or cardiovascular health is based on the techniques described herein. Can be evaluated.

  Embodiments of the present invention overcome deficiencies in conventional ST portion analysis. The RR interval and / or QT or RR interval data sets can be used to assess a patient's cardiovascular health under certain conditions. Embodiments of the present invention assess cardiovascular health in patients with varying degrees of ischemia, such as patients with coronary artery disease (“CAD”), patients with moderate ischemia, and patients with undetectable ischemia. Useful to do. A data set of RR intervals and / or QT or RR intervals can be collected during phases of changing workload and heart rate. Data collected during a recovery period after a relatively high workload and / or heart rate cycle can be analyzed as an indicator of cardiovascular health.

  More specifically, the peak heart rate of a subject during heavy work is sufficiently high so that a quasi-steady increasing period of heart rate and a subsequent recovery period are observed. The recovery period includes a decrease in the subject's heart rate by gradually reducing the work or by abruptly stopping the work of exercise. The work reduction occurs after a “cool down” phase of reducing the work. The RR interval and / or QT or RR interval data sets collected during the relatively high work and heart rate phases and during the recovery phase where subsequent work is reduced or eliminated are analyzed and the patient Can show cardiovascular health status.

  Although the present invention will be described herein primarily with reference to using QT and RR interval data sets, it will be appreciated that the present invention can be implemented in a simplified form using only RR interval data sets. Like. As used in the claims below, it is understood that the term “RR interval data set” is intended to include embodiments of both the QT-RR interval and RR interval data sets as well as the RR interval data set alone. Like. However, the case where the data set does not include the QT interval data set is specifically followed is excluded.

A first aspect of the present invention is a method for assessing cardiac ischemia in a subject to provide a measure of the cardiovascular health of the subject. The method includes the following steps:
(A) collecting a first RR interval data set from the subject during a phase in which the heart rate gradually increases to a predetermined threshold, eg, at least 130 beats / minute;
(B) collecting a second RR interval data set from the subject during a phase in which the heart rate gradually decreases;
(C) comparing the first RR interval data set and the second RR interval data set to determine a difference between the data sets;
(D) From the comparison in step (c), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the health status of the heart or cardiovascular is reduced Generating a measure of cardiac ischemia while the subject is stimulated.

  During periods of gradual increase and decrease in heart rate, the effects of sympathetic, parasympathetic, and hormonal control on the formation of hysteresis loops are sufficiently small to minimize or control to reduce ischemic changes. It can be easily detected. This maintenance is realized by the heart rate becoming sufficiently high so that the period during which the heart rate increases and decreases is a quasi-stationary period.

Another aspect of the invention is a computer program product for assessing cardiac ischemia in a subject to provide a measure of the subject's heart or cardiovascular health. The computer program product includes a computer-usable recording medium, and the computer-usable recording medium has computer readable program code means incorporated in the medium. The computer-readable program code is
(A) a computer readable program code for collecting a first RR interval data set from the subject during a phase in which the heart rate gradually increases to a predetermined threshold, eg, at least 130 beats / minute When,
(B) a computer readable program code for collecting a second RR interval data set from the subject during a phase in which the heart rate gradually decreases;
(C) a computer readable program code for comparing the first RR interval data set and the second RR interval data set to determine a difference between the data sets;
(D) From the comparison in the computer readable program code (c), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the heart or cardiovascular health And computer readable program code for generating a measure of cardiac ischemia during stimulation of the subject indicating that the subject is being reduced.

  The invention is explained in more detail in the drawings and the specification described below.

  The invention is described in more detail as follows. This description is not intended to be a detailed overview of all the various ways in which particular elements of the invention can be implemented. Numerous variations will become apparent to those skilled in the art based on the current disclosure.

Those skilled in the art will appreciate that some aspects of the invention can be embodied as methods, data processing systems, or computer program products. Accordingly, some aspects of the invention take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects.
Further, some aspects of the invention employ a form of a computer program product on a computer usable recording medium having computer readable program code means embedded in the medium. Any suitable computer readable medium may be used including, but not limited to, hard disks, CD-ROMs, optical storage devices, and magnetic storage devices.

  Several aspects of the invention are described below with reference to flowchart illustrations of methods, apparatus (systems), and computer program products. It will be understood that each block of the flowchart illustrations and combinations of blocks in the flowchart illustrations can be performed by computer program instructions. These computer program instructions are sent to a general purpose computer, special purpose computer, or other programmable data processing device processor for execution by a computer or other programmable data processing device processor. And make a device that creates a means to perform the function specified in the block of the flowchart.

  Computer program instructions capable of causing a computer or other programmable data processing device to function in a particular manner are stored in a computer readable memory, and the instructions stored in the computer readable memory are flowcharted. It is also possible to make a product including instruction means for executing the function specified in the block.

  A computer program instruction is loaded onto a computer or other programmable data processing device and a series of operational steps are performed on the computer or other programmable device to execute on the computer or other programmable device. It is also possible to create a computer-implemented process in which instructions that provide the steps to implement the functions specified in the flowchart blocks.

[1. Definition]
“Cardiac ischemia” refers to a lack or insufficient supply of blood to the myocardial region. Cardiac ischemia usually occurs when there is an arteriosclerotic occlusion of one or more coronary arteries. Atherosclerosis is a product of the lipid deposition process, resulting in the accumulation of fibrous fat or plaque that grows on the inner wall of the coronary artery. Such occlusion reduces blood flow through the arteries, and this reduction can impair the supply of oxygen to surrounding tissues while increasing physiological need, for example, increasing exercise load. Become. In later stages of cardiac ischemia (eg, where there is significant blockage of the coronary arteries), blood supply is inadequate even during myocardial rest. However, at that early stage, such ischemia is restored to its original state in the same manner as when the heart muscle returns to normal function when the oxygen supply to the heart muscle returns to normal physiological levels. Can do. Thus, ischemia that can be detected by the present invention includes temporary, chronic, and acute ischemia.

  “Exercise” as used herein refers to the subject's voluntary skeletal muscle activity, which increases the heart rate seen or elevated in steady resting state. Examples of exercise include, but are not limited to, cycling, rowing, weightlifting, walking, running, and stair climbing. These movements can be performed on a stationary device such as a treadmill or in a non-stationary environment.

  “Exercise load” or “load level” refers to the relative intensity of a particular exercise that uses a greater load or load level for a given exercise where the subject's heart rate is higher. For example, weight lifting can increase the load by increasing the amount of weight, while walking or running increases the load by increasing speed and / or increasing the slope or slope of the walking or running surface. can do.

  “Gradually increasing” and “gradually decreasing” an exercise load refers to an exercise that causes a subject to exercise under a plurality of various sequentially increasing or decreasing loads. Since the number of steps in the sequence can be infinite, the terms gradually increasing and decreasing the load include continuous load increases and decreases, respectively.

  “Hysteresis” refers to the delay of physiological effects when external conditions change.

  A “hysteresis curve” is one in which one curve reflects the response of the system to the first sequence in such a state that the heart rate gradually increases, and the other curve has a state in which the heart rate gradually decreases. A pair of curves that reflects the response of the system to the second sequence. Here, both sets of states are essentially the same, i.e., consist of the same (or nearly the same) steps, but the order of passage through time is different. A “hysteresis loop” refers to a loop formed by a pair of two continuous curves.

  An “electrocardiogram” or “ECG” is a continuous or sequential recording (or set of such recordings) of a local potential field obtained from one or more locations outside the myocardium Point to. This field is generated by the combined electrical activity of many cardiac cells (generation of action potentials). The recording electrode is implanted subcutaneously or temporarily attached to the skin surface of the subject's normal breast. An ECG recording generally includes a single lead ECG signal that indicates the potential difference between any two recording sites, including a zero or ground potential site.

  “Non-stationary intervening rest”, when used in relation to a later stage of increasing cardiac stimulation, reduces cardiac stimulation very rapidly (eg, exercise) to cause a distinct sympathetic adrenal response. It refers to the phase of time initiated by a rapid decrease in load. For this reason, an intervening resting phase that is not quasi-stationary is characterized by rapid sympathetic adrenal adjustment (described in more detail in Example 8 below) and by including an intervening resting phase, A quasi-stationary movement (or stimulation) protocol is not required (further described in Example 9 below). In contrast, “quasi-stationary intervening rest” refers to a phase of time initiated by a decrease in cardiac stimulation that does not cause a clear sympathetic adrenal response. Examples of conditions in which quasi-stationary intervening rests occur include increasing the heart rate to a sufficiently high predetermined threshold such that subsequent intervening rests are quasi-stationary.

“Present threshold heart rate” means a sufficiently high heart rate range where, for example, a decrease in exercise load or a subsequent decrease in heart rate due to removal of exercise load is quasi-stationary. Refers to that. This predetermined limit heart rate can be at least 130, 140, 150, 160 or more beats per minute. Furthermore, this predetermined limit heart rate can be defined by the following formula:
PT = (220 ± V bpm) -age
Here, PT is a predetermined limit heart rate, V is 10, 15, 20, 25 or 30, bpm is beats per minute, and age is the age of the subject. This critical heart rate corresponds to a work load between about 5, 10 or 20 W and about 30, 40 or 60 W for patients with coronary artery disease and about 80, 90 or 100 W and about 130 for patients without coronary artery disease. , 150 or 170W. This critical heart rate can also be determined by observing the amount of stress that the patient experiences during exercise so that the critical heart rate matches the degree of stress close to the maximum safe stress level.

“Quasi-steady state” means that a gradual change in the external state and / or the physiological response it causes occurs slower than any corresponding adjustment by sympathetic / parasympathetic and hormonal control. Refers to any state. _{Given a} typical time of external state variation as τ _ext , where τ _int is the earliest typical time of internal sympathetic / parasympathetic and hormonal control, the “quasi-steady state” is τ Indicates _ext >> τ _int (eg, τ _ext is at least about 2, 3, 4 or 5 times greater than τ _int ). The sudden change in exercise load can be performed in either a quasi-steady state or a non-quasi-steady state. “Rapid change that is not quasi-stationary” refers to a state opposite to a quasi-stationary state corresponding to a very fast change in the external state when compared to sympathetic / parasympathetic and hormonal control. That is, it requires that τ _ext << τ _int (eg, τ _ex t is at least about 2, 3, 4 or 5 times smaller than τ _in t). A “quasi-stationary rapid change” refers to a relatively fast change in a motion that is still quasi-stationary, for example, because a slow quasi-stationary recovery period is observed, for example when the change is preceded by a sufficiently large exercise load. Refers to that.

The “QT and RR data set” refers to a recording of the time course of an electrical signal that includes an action potential that extends through the myocardium. Any single lead ECG recording, usually referred to as a QRS complex, includes a group of three consecutive sharp shakes generated by the propagation of an anterior portion of the action potential through the ventricle. In contrast, electrical recovery of ventricular tissue is seen on the ECG as a relatively small shake known as a T wave. The time interval between cardiac cycles (ie, between the maximum of consecutive R waves) is called the RR interval, while the duration of the action potential (ie, between the start of the QRS complex and the end of the subsequent T wave). Is called the QT interval. Other definitions of these intervals can also be used in a similar sense within the framework of the present invention. For example, the RR interval is defined as the time between any two similar points, such as similar inflection points on two consecutive R waves, or any other method of measuring the cardiac cycle length. be able to. The QT interval can be defined as the time interval between the peak of the Q wave and the peak of the T wave. The QT interval is any time interval between the start point (or center) of the Q wave and the end point of the subsequent T wave defined as a point on the time axis (baseline), or the duration of the action potential. It can also be defined as other methods. At the end of the T wave, the QT interval intersects the linear extrapolation of the falling branch of the T wave and begins at that inflection point. A set of durations of such intervals, ordered simultaneously at the start or end time accumulated on a basis between beats or on the basis of any given beat sampling rate, is the corresponding QT and RR interval data. Form a set. For this reason, the QT and RR interval data sets are stored in an array {T _QT, 1, T _{QT, 2,.} . . , T _{QT, n} } and {t ₁ , t _{2 ,.} . . , T _n } and an array {T _{RR, 1} , T _{RR, 2,.} . . , T _{RR, n} } and {t ₁ , t ₂ ,. . . , T _n } (the array {t ₁ , t ₂ ,..., T _n } may or may not exactly match a similar array in the QT data set).

は、Ｃ［ａ，ｂ］から任意の絶対連続関数Ｆに対して積分として定義される（スティルチェス積分）。

区間［ａ，ｂ］上で単調な関数Ｆに関して、その変化量は単に｜Ｆ（ａ）−Ｆ（ｂ）｜となる。関数Ｆ（ｔ）が交互的な最大値および最小値を有する場合、Ｆの全変化量は、単調性の間隔上のその変化量の合計である。例えば、最小値および最大値の点がｘ₁＝ａ，ｘ₂，ｘ₃，．．．，ｘ_k＝ｂの場合は、以下の式のようになる。

In the following definition, C [a, b] indicates a set of continuous functions f (t) on the interval [a, b]. {T _i }, i = 1, 2,. . . , N denote the set of points from [a, b], ie {t _i } = {t _i : a ≦ t _i ≦ b, i = 1, 2,. . . , N} and {f (t _i )}. Here, fεC [a, b] indicates a set of values of the function f at the point {t _i }. In matrix operation, the quantity τ = {t _i }, y = {f (t _i )} is treated as a column vector. E _N represents an N-dimensional metric space having a distance R _N (x, y), x, yεE _N. (R _N (x, y) is said to be the distance between points x and y.) (Total) Change

Is defined as an integral for any absolute continuous function F from C [a, b] (Still chess integral).

Regarding the monotonous function F on the interval [a, b], the amount of change is simply | F (a) −F (b) |. If the function F (t) has alternating maximum and minimum values, the total change in F is the sum of its changes over the monotonic interval. For example, if the minimum and maximum points are x ₁ = a, x ₂ , x ₃ ,. . . , X _k = b, the following equation is obtained.

をＣ［ａ，ｂ］のサブセットとする。

の場合、連続した関数

は、データーセット｛ｘ_i，ｔ_i｝（ｉ＝１，２，．．．，Ｎ）に対する距離Ｒ_Nに関してクラス

の「（最良の）フィット（または最良のフィッティング）関数」と呼ばれる。この時、Ｒ_Nの最小値はフィットのエラーと呼ばれる。

からの関数ｆ（ｔ）は、試行関数と呼ばれる。 “Fitting” (best fitting):

Be a subset of C [a, b].

A continuous function

Is the class with respect to the distance R _N for the data set {x _i , t _i } (i = 1, 2,..., N).

Called the “(best) fit (or best fitting) function”. In this case, the minimum value of R _N is called the error of the fit.

The function f (t) from is called the trial function.

上で実行される。 In most cases, E _N means Euclidean space with Euclidean distance. At this time, the error _RN becomes a well-known root mean square error. The fit usually implies a specific parameterization of the trial function and / or constraints as a requirement that the trial function pass through a given point and / or have a given value slope at a given point. Subset

Executed above.

および、

が成立する場合、関数ｇ（ｔ）よりも平滑である。ここで、素数は時間導関数を表し、厳密な不等式は関係式（式Ｄ．４）および（式Ｄ．５）の少なくとも１つを維持する。 “Smoother function (smoothness comparison)”:
Let f (t) and g (t) be functions from C [a, b] that have absolute continuous derivatives over this interval. The function f (t) is

and,

Is true, the function is smoother than the function g (t). Here, the prime number represents a time derivative, and the strict inequality maintains at least one of the relational expressions (formula D.4) and (formula D.5).

"Smooser Set":
The set {x _i , t _i } (i = 1, 2,..., N) is smoother than the set {x ′ _j , t ′ _j } (j = 1, 2,..., N ′). It is. This is the case when the former can be fitted to the same class of smoother functions f (t) with the same or less error than the latter.

"Data set smoothing":
Data set (x, t) = {x _i , t _i } (i = 1, 2,..., N ₀ ) in the following form y = A · x, τ = B · t (formula D.6)
(Linear) transformation to another set (y, τ) ≡ {y _j , τ _j } (j = 1, 2,..., N ₁ ) in the case where the latter set is smoother than the former Is called “smoothing”. Here, A and B are N ₁ × N ₀ matrices. {Y _j , τ _j } can be called a smoothed set.

ここで、ρ（τ，Ｔ）は、Ω上の非負の（重み）関数である。 “Measure of closed domain”:
Let Ω be a singly connected region on a plane (τ, T) having a boundary formed by a simple (ie, no self-intersection) continuous curve. The scale M of such a region Ω on the plane (τ, T) is defined as the Riemann integral as follows:

Here, ρ (τ, T) is a non-negative (weight) function on Ω.

If ρ (τ, T) ≡1, the scale M of the region is coincident with its area A, and if ρ (τ, T) ≡1 / τ ² , the scale M is the transformed plane (f, T Note that it means the area A ′ of the upper region Ω ′). Here, f = 1 / τ can be understood as a heart rate since the quantity τ has the meaning of the RR interval. (Region Ω ′ is an image of region Ω in mapping (τ, T) → (1 / τ, T).)

[2. Dispersion / recovery curve]
FIG. 1 shows a periodic action potential (AP, upper graph 20) generated in the myocardium and summed over its entire volume, and an electrocardiogram (ECG, lower graph 21) generated on the body surface. The correspondence between the temporal phases with the recorded electrical signal is shown. This figure shows two regular heart cycles. A QRS composite wave is formed during the action potential rising movement. The composite wave consists of three waveforms, Q, R and S, marked on the lower panel. The recovery phase of the action potential is characterized by its dip on the AP plot and the T wave on the ECG plot. It can be seen that the duration of the action potential is adequately indicated by the time between the Q wave and the T wave and is conventionally defined as the QT interval measured from the beginning of the Q wave to the end of the subsequent T wave. The time between successive R waves (RR interval) indicates the duration of the cardiac cycle, while its reciprocal value indicates the corresponding instantaneous heart rate.

  FIG. 2 shows the main aspects of the periodic action potential propagation through the heart tissue and the formation of the corresponding composite wave dispersion-recovery curve. Tissue can be thought of as a continuous medium, and the propagation process can be thought of as a repetition at each media point of the continuous phase of excitement and recovery. The former aspect is characterized by rapid growth of local membrane potential (depolarization) and the latter is characterized by its return to negative rest values (repolarization). The excitement phase includes a very fast (about 0.1 ms) decrease in the excitatory threshold and subsequent rapid inward sodium current generation (about 1 ms) that causes the action potential to rise. Next, during the intermediate plateau phase (about 200 ms), the sodium current is inactivated and the calcium and potassium currents are reduced while the membrane is temporarily not excited (ie, the threshold is high). Occur. During the next recovery phase (approximately 100 ms), the membrane re-excites (excitement threshold is lowered) as the potassium current repolarizes the membrane.

A complicated description of the large number of ion currents involved in the process can be avoided if the process is handled directly by the local membrane potential u and the local excitation threshold v. Such a mathematical description is called the CSC model and was developed by Chernyak, Starobin, and Cohen (Phys. Rev. Lett., 80, pp.5675-5678, 1998), with two reactions in panel A − Shown as a set of diffusion (RD) equations. The left hand side of the first equation describes the local accumulation of charge on the membrane, the first term on the right hand side describes the ohmic coupling between adjacent points of the medium, and the term i (u, v) is the membrane. It shows transmembrane current as a function of potential and changing excitatory threshold (ε is a small constant, slow recovery rate to fast excitability rate). Periodic solutions (wave trains) can be found analytically for some specific functions i (u, v) and g (u, v). The wave train shown in panel B was calculated for g (u, v) = ζu + v _r −v. Where ζ and v _r are appropriately selected constants (v _r means the initial excitability threshold and is the main determinant of media excitability). The function i (u, v) is composed of two linear parts, one for the area u <v below the threshold and one for the area u> v above the threshold. Selected. That is, i (u, v) = λ _r u when u ≦ v, and i (u, v) = λ _ex (u−u _ex ) when u> v. Where λ _r and λ _ex are the membrane chord conductances in the resting state (u = 0) and the excited state (u = u _ex ), respectively. The rest state u = 0 is the origin of the potential scale. The inventors used units of λ _ex = 1 and u _ex = 1. (For details, see Chernyak and Starobin's Critical Reviews in Biomed. Eng. 27, 359-414 (1999).)

A highly excitable medium corresponding to a well-conducted tissue causes a faster and more active action potential with a longer action potential duration (APD). This state also means that longer-lasting excitement propagates faster. Similarly, high frequency wave trains are slow to propagate due to less time for the medium to recover from the preceding excitement and thus less effective excitability. These are very general features unique to the CSC model. Physically, the relationship between the wave velocity c and its frequency f or its period T = 1 / f is called a dispersion relationship. In the CSC model, this dispersion relation can be obtained in the explicit form T = F _T (c). Where F _T is a well-known function of c and media parameters. With the CSC model, the relationship between propagation speed and ADP T _AP can be found in the explicit form T _AP = F _AP (c). This indicates the recovery characteristics of the medium. In the medical literature, the recovery curve is _TAP vs. diastolic interval _TDI , which makes another very similar physical representation. As shown in panel C (FIG. 2), a pair of dispersion and recovery relations {T = F _T (c), T _AP = F _AP (c)} is expressed as one curve on the (T, T _AP ) plane. It can be considered as a parameter display. Such a curve (relational expression) is called a composite dispersion-recovery curve (relational expression) and is obtained directly by measuring a QT or RR interval data set from an experimental ECG recording and plotting T _QT vs. T _RR. be able to. The condition that the {T _QT , T _RR } data set obtained from the experiment actually shows a composite dispersion-recovery relationship requires that the data be collected under quasi-steady state. Understanding this fact is an important discovery for the present invention.

[3. Test method]
The method of the present invention is primarily aimed at testing subjects. Virtually any subject, including male, female, youth, youth, adult and elderly subjects, can be tested by the method of the present invention. This method can be performed as an initial screening test for subjects who do not use a sufficient medical history or previous records, and can be performed repeatedly on the same subjects (especially comparisons of individual heart health over time). If quantitative signs are required), the effects and / or effects of the intervening events and / or intervening treatments on the subject can be assessed during the study period.

As mentioned above, the method of the present invention generally comprises:
(A) collecting a first RR interval data set from the subject during a phase in which the heart rate gradually increases to a predetermined threshold of at least 130 beats / minute;
(B) collecting a second RR interval data set from the subject during a phase in which the heart rate gradually decreases;
(C) comparing the first RR interval data set and the second RR interval data set to determine a difference between the data sets;
(D) From the comparison in step (c), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the health status of the heart or cardiovascular is reduced Generating a measure of cardiac ischemia while the subject is stimulated.

  The phase in which the heart rate gradually increases and decreases is that during both periods, almost or approximately the same stimulus is applied to the heart, not the effect of external control as measured by the means of the present invention, It is performed in such a way that it is maintained by the peripheral nervous system and the hormonal control system, as is the effect of blood. This method can be performed by a number of techniques, including techniques that perform two successive steps of gradually increasing and gradually decreasing exercise load (or average heart rate).

  The duration of gradually increasing exercise load (or increasing average heart rate) and gradually decreasing exercise load (or decreasing average heart rate) is the same or different. Also good. In general, the duration of each stage is at least 3, 5, 8 or 10 minutes or more. At the same time, the duration of the two stages can be from about 6, 10, 16 or 20 minutes to about 30, 40 or 60 minutes or more. The two phases are preferably carried out in succession in time, i.e. one phase is almost immediately after the other, without the intervening resting phase. Alternatively, the two phases can be performed with a “plateau” phase (eg, 1 to 5 minutes) intervening in time. During the plateau phase, the cardiac stimulation or exercise load remains approximately constant until the beginning of the phase of reducing the load. In certain embodiments described later where the subject has a detectable coronary artery disease, the step of gradually reducing the exercise load is performed during a phase after a quasi-stationary sudden movement stop Can be considered.

  In the exercise protocol, the set of loading steps during the phase where the heart rate is increased or decreased may be the same or different. For example, the peak load at each stage may be the same or different, and the minimum load at each stage may be the same or different. In general, each stage consists of at least two or three different load levels in ascending or descending order depending on the stage. A relatively high load level such that the heart rate is relatively high can be used. The peak or predetermined threshold heart rate is high enough so that subsequent periods of resting or decreasing exercise load are quasi-stationary. Embodiments of the present invention can be used to diagnose or test various levels of ischemia, including patients whose ischemia cannot be detected, patients with moderate ischemia, and patients with coronary artery disease.

  The predetermined limit heart rate corresponds to the amount of work that varies for different patients depending on the overall health of the body and the health of the heart. For example, for exercised or trained subjects, in the first or ascending phase, the first load level is selected such that the subject requires a power output of 60 to 100 or 150 watts and the intermediate load level is The subject is selected to require a power output of 100 to 150 or 200 watts, and the third load level is selected such that the subject requires a power output of 200 to 300 or 450 watts or more. For the second or descending phase, the first load level is selected such that the subject requires a power output of 200 to 300 or 450 watts or more, and the intermediate or second load level is 100 to 150 or The power output is selected to require 200 watts, and the third load level is selected such that the subject requires a power output of 60 to 100 or 150 watts. Different load levels can be included as needed before, after, or between all the above load levels, and adjustment between load levels can be any suitable method, including stepped or continuous methods Can be done. An increased heart rate to a predetermined limit heart rate can also be achieved by maintaining one sustained load level for a period sufficient to reach the limit heart rate.

  In yet another embodiment, for an average subject or a subject with a cardiovascular history, in the first or rising phase, the first load level requires the subject to have a power output of 40 to 75 or 100 watts. The intermediate load level is selected such that the subject requires a power output of 75 to 100 or 150 watts, and the third load level requires a power output of 125 to 200 or 300 watts or more. Is selected. For the second or descending phase, the first load level is selected such that the subject requires a power output of 125 to 200 or 300 watts or more, and the intermediate or second load level is 75 to 100 or The 150 watt power output is selected and the third load level is selected so that the subject requires a 40 to 75 or 100 watt power output. As described above, another load level can be entered before, after, or between all the aforementioned load levels if necessary, and adjustment between load levels can be any step including a stepped or continuous method It can be done by any suitable method. The predetermined limit heart rate can also be achieved by continuing the exercise of one power output in sufficient time.

  By subjecting the patient to a predetermined scheduled stimulus, the heart rate can gradually increase and decrease. For example, based on a predetermined program or schedule, the patient may gradually increase and gradually decrease exercise load, or gradually increase electrical or pharmacological stimulation and gradually decrease electrical or pharmacology It can be applied to typical stimuli. There is no actual heart rate feedback from the patient for such a predetermined schedule. Alternatively, the patient's heart rate can be gradually increased and decreased in response to the actual heart rate data collected by simultaneously monitoring the patient. Such a system is a feedback system. For example, monitor the patient's heart rate during the test and adjust the exercise load (speed and / or tilt in the case of a treadmill) so that the heart rate changes in the manner prescribed during both test phases can do. Load monitoring and control can be accomplished by a computer or other control system using a simple control program and an output panel connected to the control system and exercise device that generates analog signals to the exercise device. One advantage of such a feedback system is that the control system (if necessary) indicates that the heart rate increases approximately linearly during the first phase and decreases approximately linearly during the second phase. ) It can be done reliably.

  The generating step (d) generates curves from the data set (with or without actually displaying them), and then (i) a measure of the area between the hysteresis curves (eg, area) If the scale is large, evaluate directly or indirectly (eg as defined in integration theory) to indicate that the subject has high cardiac ischemia, and (ii) the shape of the curve (eg , Slope or derivative thereof), or (iii) (i) and (ii) are combined directly or indirectly to indicate that the subject has a large cardiac ischemia if the difference in shape is large Can be performed by any suitable means. A specific example is given in Example 4 below.

  The method of the invention comprises (e) a measure of cardiac ischemia during exercise and at least one reference value (e.g. mean value, median value for quantitative display from a population or subpopulation of individuals or And (f) generating at least one quantitative indication of the subject's cardiovascular health from the comparison of step (e). Any such quantitative indication may be used to assess the potential risk of experiencing a cardiac event related to future ischemia such as myocardial infarction or ventricular tachycardia temporarily (eg, ) Generated or specifically generated for monitoring the time course of a subject, either for prescribed cardiovascular treatment or simply as a continuous monitoring of the physical condition in which the subject recovers or worsens ( Again, a specific example is given in Example 4 below). In such cases, steps (a) through (f) above are repeated in at least one individual case to assess the effectiveness or course of cardiovascular treatment of the subject. A decrease in the difference between the data sets from before the treatment to after the treatment or over time indicates that the subject's heart health has improved from the cardiovascular treatment. Aerobic exercises, strength building, diet changes, nutritional supplements, weight loss, smoking cessation, stress reduction, drug treatment (including gene therapy), surgical treatment (bypass, balloon angioplasty, catheter ablation, open heart and open heart surgery) Any suitable cardiovascular therapy including, but not limited to, and combinations thereof.

  The treatment or intervention may be approved or experimental. In the latter case, the present invention is carried out in the context of a clinical trial of experimental treatment, which is conducted before and after treatment (and / or during treatment) for the purpose of determining the effect of the proposed treatment.

[4. Test equipment]
FIG. 3 provides an embodiment of an apparatus for collecting, processing and analyzing data according to the present invention. An electrocardiogram is recorded by the ECG recorder 30 via a lead placed on the subject's body. The ECG recorder can be, for example, a standard multi-lead Holter recorder or any other suitable recorder. The analog-to-digital converter 31 digitizes the signals recorded by the ECG recorder and transfers these signals to a personal computer 32 or other computer or central processing unit via standard external input / output ports. The digitized ECG data is then processed by standard computer-based waveform analyzer software. Next, a composite dispersion-recovery curve and an indication or other quantitative measure of heart or cardiovascular health relative to the presence, absence or extent of cardiac ischemia can be used in software, hardware, or hardware. And automatically calculated in the computer by programs (eg, Basic, Fortran, C ++, etc.) that run internally as both software and software.

4A and 4B illustrate the main steps of digitized data processing to perform analysis of QT or RR data sets collected from subjects during a reciprocal quasi-stationary change in physiological conditions. . The first four steps of FIGS. 4A and 4B are substantially the same. Digitized data collected from a multi-lead recorder is stored 40a, 40b in a computer memory as a data array for each lead. The size of each data array is determined by the duration of the rise and fall phases of the heart rate and the sampling rate used by the waveform analyzer that processes the incoming digitized ECG signal. The waveform analyzer software first detects 41a, 41b the main characteristic waves (Q, R, S and T waves) of the ECG signal at each particular lead. Next, at each ECG lead, the software determines the time interval 42a, 42b between successive R waves and between the start point of the Q wave and the end point of the T wave. Using these reference points, the software calculates heart rate and RR and QT intervals. Next, the application part of the software sorts 43a, 43b intervals for the stages in which the heart rate rises and falls. The next two steps are performed in one of two alternative methods shown in FIGS. 4A and 4B, respectively. In the fifth step shown in FIG. 4A, a QT interval vs. RR interval is set by the software application unit so that the heart rate performed by a gentle reciprocal change in a physiological state such as exercise, pharmacological / electrical stimulation, Step 44a is displayed separately for the rising and lowering stages. The same part of the software performs the next step 45a. This step 45a is smoothed, filtered, or data-driven using an exponential function or any other suitable function to obtain a sufficiently smooth curve T _QT = F (T _RR ) for each stage. This is the step of fitting. Another method for the last two steps shown in FIG. 4B is that the software application uses an exponential function or any other suitable function to average the QT interval as a function of time for the two stages. And / or filtering and / or fitting, and processing the RR interval data set in a similar manner, two sufficiently smooth curves T _QT = F _QT (each containing branches where the heart rate rises and falls t) and T _RR = F _RR (t) need to be generated 44b. In the next step 45b, the application part of the software removes time using this parameter display and generates and plots a sufficiently smooth hysteresis loop T _QT = F (T _RR ). The next steps shown in FIGS. 4A and 4B are similar in this case. The next steps 46a, 46b performed by the application part of the software are appropriate, such as a vertical straight line or a line connecting the start and end points, to create a closed hysteresis loop on the (T _QT , T _RR ) plane. An interconnecting line or a partially connecting line can be used to close the hysteresis loop of the two branches. In the next steps 47a, 47b, the application software evaluates the appropriate size of the area inside the closed hysteresis loop for each ECG lead. Size is a generalization of the concept of area, as defined in mathematical integration theory, and includes appropriate weight functions that increase or decrease the contribution of different points of the region to this size. Yes. The final step 48a, 48b of data processing for each ECG lead is that the application software computes the exponent by appropriately rolling this magnitude or any monotonic function of this magnitude. The magnitude itself along with the index reflects the severity of both ischemia caused by exercise as well as a predisposition to local ischemia. This predisposition to local ischemia is reflected in several properties of the shape of the measured composite dispersion-recovery curve. The results of all the aforementioned signal processing steps can be used to quantitatively assess the cardiac ischemia of the particular individual being tested and, optionally, the cardiovascular health status at the same time.

(T _QT, T _RR) instead of using the plane, similar data processing method, (T _QT, f _RR), such as (T _QT, T _RR) nondegenerate transform plane (non-degenerate transformation) Can be carried out in the same way on any plane obtained by. Here, f _RR = 1 / T _RR is a heart rate. Some or all of such transformations can be incorporated into an appropriate definition of the size.

[5. Conditions under which a sudden stopping motion protocol can be considered a quasi-stationary protocol
In general, the duration of each stage of gradually increasing or decreasing the quasi-steady movement protocol is at least 2, 5, 8 or 10 minutes. The duration of each stage consists of an average peak load heart rate value (about 120-130 to about 150 or 160 beats / minute) and an average resting heart rate value (about 50 or 60 to about 70 or 80 beats / minute). ) Is typically an order of magnitude longer than the average duration of heart rate adjustment (approximately 1 minute) during a sudden movement stop.

The short approximately 1 minute heart rate adjustment period after a sudden quasi-stationary sudden movement stop is generally relatively low for a healthy person and a relatively low exercise load corresponding to a power output not exceeding, for example, 75 W (watts). Note that this is only done for patients with moderate or moderate coronary artery disease (CAD). However, for an ischemic patient with coronary artery disease or a normal person without coronary artery disease who has received a relatively large exercise load with a relatively high heart rate, the duration of the same adjustment period is, for example, 5 It can be 8 or 10 minutes or longer. The critical heart rate that becomes quasi-stationary with increasing adjustment period can be defined by the following formula:
PT = (220 ± V bpm) -age
Here, PT is a predetermined limit heart rate, V is 15, 20, or 25, bpm is beats per minute, and age is the age of the subject. This limit heart rate is said to be a heart rate close to a predetermined maximum heart rate. In these cases, heart rate variability is small (as described in Example 9 below) and during the first 1 or 2 minutes after exercise, no further exercise load or for example 25W Even when subjected to the following light loads, individuals show that they continue to be under great physical stress even after a sudden cessation of movement. Under the condition of a relatively large exercise load, the recovery portion of the QT / RR hysteresis loop can be considered as a quasi-stationary loop, as is the overall loop finally formed after a sudden exercise stop. In fact, the heart rate slowly recovers to its pre-exercise level, which takes 2, 5, 8, 10 minutes or more, and the heart rate variability is so small that it defines the basis of a gentle exercise protocol Is still satisfied (see Example 9 below).

  Thus, if the work of exercise is sufficiently large, a sudden movement stop does not prevent the quasi-stationary end of the descending HR recovery phase in the method according to embodiments of the present invention. The reason is that, due to the long recovery of HR, the QT / RR hysteresis can still be treated as quasi-stationary, and despite the fact that the descending phase of the motion protocol is not complete, the hysteresis loop This is because the size can be evaluated by an ordinary method.

  In general, patients suffering from overt coronary artery disease as determined by physical exhaustion, shortness of breath, chest pain and / or some other clinical condition, even if the power output is as low as about 20 watts, Note that it is not possible to exercise 3, 5, 8 or 10 minutes or longer. The slow recovery process after such a sudden cessation of this type of patient meets the definition of a quasi-stationary process as given in this application (see, eg, Example 9). Note that patients with moderate coronary artery disease can perform longer exercises of 20 minutes or longer and can receive significantly higher exercise loads in the range of 50 to 300 Watts.

  The present invention will be described in more detail in the following examples, but is not limited to these examples.

[Example 1]
<Test equipment>
The same test equipment as FIG. 3 was assembled. Lead-Lok Holter / Stress Test Electrodes LL510 (LEAD-LOK, INC .; 500 Airport Way, POBox L, Sandpoint, ID, USA 83864) with an ECG placed on the subject's body based on the manufacturer's instructions Recorded by 12 RZ152PM12 Digital ECG Holter Recorder (ROZINN ELECTRONICS, INC .; 71-22 Myrtle Av., Glendale, New York, USA 11385-7254) via 12 conductors. Digital ECG data is stored in a personal computer (Dell Dimension XPS T500MHz / Windows® 98) using a PC 700 flash card reader and a 40 MB flash card (RZFC40) (both from Rozinn Electronics, Inc.). ). The Holter waveform analysis software for Windows® (4.0.25) is installed on the computer. This software is used to process the data with standard computer-based waveform analyzer software. Next, a display providing quantitative characteristics for the composite dispersion-recovery curve and the degree of cardiac ischemia is calculated either manually or automatically in the computer by a program executed in the Fortran 90.

  Experimental data was collected during the exercise protocol programmed in the Landice L7 Executive Treadmill (Landice Treadmills; 111 Canfield Av., Randolph, NJ 07869). This programmed protocol contains 20 stepped intervals with a constant exercise load of 48 seconds to 1.5 minutes each duration. Throughout these intervals, the overall duration varied from 16 minutes to 30 minutes, creating a phase of gradually increasing and decreasing the exercise duration of two equal durations. For each stage, the treadmill belt speed and height varies from 2.4 km / hr (1.5 miles / hr) to 8.8 km / hr, depending on the age and health of the subject. (5.5 miles per hour) and the treadmill height changed from 1 to 10 steps.

[Examples 2 to 6]
<Study of human hysteresis curve>
These examples show the QT-RR interval hysteresis caused by quasi-stationary ischemia in a number of different subjects. These data demonstrate that this method has high sensitivity and high resolution.

[Examples 2-3]
<Hysteresis curve of healthy male subjects of different ages>
These examples were performed on two male subjects using the apparatus and method described in Example 1 above. Referring to FIG. 5, it can be easily seen that there is a significant difference in the area of hysteresis between two generally healthy male subjects of different ages. These subjects (23 and 47) exercised on a treadmill according to a quasi-steady 30 minute protocol that gradually increased and decreased exercise load. Here, squares and circles (thick lines) indicate a hysteresis loop of a 23 year old subject, and diamonds and triangles (thin lines) correspond to a large loop of a 47 year old subject. The fitting curve is obtained using a cubic polynomial function. The beat sampling rate at which the waveform analyzer determines the QT and RR intervals is one sample per minute. None of the subjects had a drop in the ECG-ST segment due to conventional ischemia. However, the method of the present invention allows the observation of hysteresis due to ischemia with sufficient resolution within the range below the conventional threshold of ischemic events, and quantifies the hysteresis between the two subjects. Can be distinguished from each other.

<Examples 4 to 5>
<Hysteresis curve of subjects with ST segment drop or myocardial infarction history (Prior Cardiac Infarction)>
These examples were performed on two 55 year old male subjects using the apparatus and method described in Example 1 above. FIG. 6 shows the quasi-stationary QT-RR interval hysteresis for male subjects. Curves consistent with squares and empty circles relate to the first subject and indicate cases of cardiac ischemia that can also be detected by conventional ECG-ST segment descent techniques. Curves consistent with diamonds and triangles are associated with another subject, a subject who has previously experienced myocardial infarction. These subjects exercised on a treadmill according to a quasi-steady 20 minute protocol that gradually increased and decreased exercise load. The fitting curve is obtained using a cubic polynomial function. These cases are due to the method of the present invention, (1) the difference between levels of ischemia that can be detected by the conventional ST drop method, and (2) below the threshold for the conventional method, It demonstrates that differences between low levels of ischemia (shown in FIG. 5) that cannot be detected can be resolved and quantitatively characterized. The level of ischemia due to exercise reported in FIG. 5 is significantly lower than the level shown in FIG. This fact shows that the resolution of the conventional ST descent method is insufficient compared with the method of the present invention.

[Example 6]
<Hysteresis curve by the same subject before and after regular exercise system>
This example was implemented using the apparatus and method described in Example 1 above. FIG. 7 provides an example of quasi-stationary hysteresis before and after a 55 year old male subject performs regular aerobic exercises. Both experiments were performed based on the same quasi-stationary 20 minute protocol that gradually increased and decreased the exercise load. The fitting curve is obtained using a cubic polynomial function. The first test shows a cardiac ischemic event resulting from overt motion created at approximately peak levels of exercise load detected by both the method of the present invention and the conventional ECG-ST descent method. The maximum heart rate reached during this first test (before starting the regular exercise system) was equal to 146. After a regular course of exercise, subjects improved their cardiovascular health. This can only be roughly evaluated qualitatively in the past by comparing peak heart rates. In fact, the maximum heart rate at peak exercise load when moving from the first experiment to the second experiment decreased from 146 to 122, decreasing by 16.4%. Although the conventional ST segment method also shows no ST drop, it did not provide any quantification of such improvement because the extent of this ischemia was below the method's threshold. Unlike such conventional methods, the method of the present invention provided just the above quantification. Applying the present invention, the curve of FIG. 7 created from the second experiment shows that the area of hysteresis in the quasi-stationary QT-RR interval has been significantly reduced from the first experiment, and such hysteresis. The loop indicates that there is still ischemia due to some level of movement. It is also shown that the observed change in the shape of the composite dispersion-recovery curve is improved. The reason is that these curves are low curves in FIG. 2 from flat curves similar to the flat curves in FIG. 2 (low excitability and high threshold, vr = 0.3 to 0.35). This is because it has changed to a healthier (less ischemic), more convex curve similar to (vr = 0.2-0.25). Thus, because of the high sensitivity and resolution of the method of the present invention, to assess minute changes in the level of cardiac ischemia, indicating changes in cardiovascular health treated by conventional cardiovascular interventions FIG. 7 demonstrates that it can be used.

[Example 7]
<Calculation for quantitative display of cardiovascular health>
This example was performed using the data obtained in Examples 2-6 above. FIG. 8 shows a comparative analysis of cardiovascular health based on ischemia assessment according to the method of the present invention. In this example, a display of cardiovascular health status (in this application, the heart ischemic index, abbreviated as “CII”) was designed. It is the area of the hysteresis loop of the quasi-stationary QT-RR interval, normalized by dividing it by the product (T _{RR, max} -T _{RR, min} ) (T _{QT, max} -T _{QT, min} ) Defined as S. For each particular subject, this factor is an area for individual differences within the actual range of QT and RR intervals that occur during testing under a quasi-stationary treadmill exercise protocol. to correct. The inventors determined the maximum and minimum CII among the 14 exercise test samples and normalized index <CII> = (CII−CII _min ) / (CII _max −CII _min ) varying from 0 to 1 Derived. The change in <CII> in various subjects is due to the method of the present invention in which heart and cardiovascular health in areas where the conventional ST drop method is below threshold and ischemia due to any exercise cannot be detected. It shows that various levels can be decomposed and characterized quantitatively. Thus, unlike ischemic assessment with coarse conventional ST segment depression, the method of the present invention is very accurate assessment and monitoring for small variations in cardiac ischemia and related cardiac or cardiovascular health changes. I will provide a.

[Example 8]
<Explanation of rapid sympathetic adrenal transients>
FIG. 9 shows typical rapid sympathetic / parasympathetic versus QT intervals (panels A, C) and RR intervals (panels B, D) for a quasi-stationary rapid stop after 10 minutes of exercise that increases exercise load. Shows nerve and hormone regulation. All panels show the time variation of the QT / RR interval obtained from the right anterior chest lead V3 of the 12 lead multi-lead ECG. The sampling rate at which the waveform analyzer measured the QT and RR intervals was equal to 15 samples / minute. The subject (47-year-old man) rested for the first 10 minutes and then began exercise (10 minutes) with a gradual increase in exercise load (left side of panel A, B-RR and QT minimum). Next, at the peak of exercise load (heart rate is about 120 beats / min), the subject descends from the treadmill and eliminates applying the earliest RR and QT intervals for a sudden stop of exercise load. did. The subject rested long enough (13 minutes) to ensure that the QT and RR intervals were on average steady-state after exercise. Panels C and D demonstrate that the maximum rate of change of QT and RR intervals occurred immediately after a sudden stop in exercise load. These rates are about 0.015 s / min for QT intervals while changing from 0.28 s to 0.295 s, and about RR intervals for changing from 0.45 s to 0.6 s. 0.15 s / min. Based on the experiments described above, a definition for "rapid sympathetic adrenal and hormonal transients" or "rapid autonomic nervous system and hormonal transients" can be given.

  Note that the maximum exercise load of Example 8 is not sufficient to cause the subject to make a quasi-stationary sudden stop. That is, this exercise load does not cause sufficient stress that requires an extended recovery period as observed in Examples 9 and 10. A rapid transient due to autonomic nervous system and hormonal control refers to a transient with a rate of 0.15 s / min corresponding to a rate of change of about 25 beats / min of the heart rate for the RR interval, and QT interval Is a transient phenomenon at a speed of 0.02 s / min. Or this rapid transient refers to the faster rate of change in the RR / QT interval in response to a very rapid change (stop or increase) in exercise load (or other cardiac stimulation). A very rapid change in exercise load is defined herein as a change in load that causes a rapid change in the RR / QT interval, comparable to the magnitude of the entire range from the peak of exercise to the steady average rest value.

[Example 9]
<Description of quasi-stationary exercise protocol with low heart rate>
FIG. 10 shows a typical slow (quasi-stationary) QT measured while gradually increasing and decreasing exercise load in the right anterior chest V3 lead of a 12 lead electrocardiogram recording. The adjustment of the spacing (panel A) and the RR spacing (panel B) is illustrated. Sampling was at 15 QT and RR intervals per minute. Male subjects exercised between two consecutive long 10-minute phases that gradually increased and decreased exercise load. Both QT and RR intervals gradually approach a minimum at approximately peak exercise load (peak heart rate is approximately 120 beats / minute) and then gradually return to a slightly lower level than the rest value before the first exercise. It was. The expansion of the QT and RR intervals was well approximated by an exponential fitting curve shown in gray in panels A and B. The range of round trip time variation of the QT-RR interval is 0.34 s to 0.27 s to 0.33 s (average rate of change is about 0.005 s / min) and 0.79 s, respectively, for the QT and RR intervals. -0.47 s-0.67 s (average rate of change was about 0.032 s / min or about 6 beats / min). The observed standard deviation σ of the root mean square of QT and RR intervals, shown as black dots in both panels from the exponential fitting, is between the corresponding peak value and rest value during the entire test. It was an order of magnitude smaller than the average difference. These deviations were respectively σ≈0.003 s for the QT interval and σ≈0.03 s for the RR interval. According to FIG. 9 (Panels C and D), this type of small perturbation can increase and decrease earlier than 10 s when combined with a sudden heart rate change due to physiological fluctuations or discontinuities in exercise. is there. That time is 60 times shorter than the duration of one gradual (rising and falling) phase of the exercise protocol. Due to such significant differences between the gradual change in the QT / RR interval and the amplitude of the rapid heart rate variability and between the time constants, these functions can be achieved by a highly accurate function such as an appropriate smoothing exponential function. The variation over time can be averaged to fit the change in duration of the QT / RR protocol. A simultaneous fitting procedure (panels A and B) determines an algorithm that removes the time dependence of the parameters from both measured QT or RR datasets and considers the QT interval for each motion stage as a monotonic function. Make it possible.

  For a mild, low load, low heart rate and quasi-stationary protocol useful for patients with coronary artery disease, see co-pending US patent application Ser. No. 10 / 036,005 filed Dec. 26, 2001. Explained

[Example 10]
<Explanation of quasi-stationary exercise protocol with high heart rate>
FIGS. 21-24 are RR interval data sets for patients with and without coronary artery disease. FIGS. 25-26 illustrate the formation of hysteresis loops for patients with and without coronary artery disease. QT and RR interval data were collected from each subject during the phase in which the heart rate gradually increased to a predetermined limit heart rate and during the phase in which the heart rate gradually decreased. The step of gradually decreasing the heart rate follows the step of reducing exercise as shown.

  FIG. 21 is a data set of the RR interval of a 60-year-old coronary artery disease in which 4.5 minutes of work was performed before a quasi-stationary sudden movement stop with a maximum work of 28 W (Watt). FIG. 22 is an RR interval data set for the first 1.5 minute recovery after a patient with the same coronary artery disease as in FIG. Heart rate recovery occurs at a quasi-stationary rate of 9 beats / minute. FIG. 25 shows the formation of a hysteresis loop for the same patient using QT and RR data. The upper and lower panels show raw data and smoothed data, respectively.

  FIG. 23 shows 50 years of age without coronary artery disease after 14 minutes of work before almost suddenly stopping exercise after 2W low recovery work of 2W after maximum work of 130W RR interval data set for subjects. FIG. 24 shows the first 1.5 after the 50-year-old subject, as in FIG. 23, stopped almost abruptly after a low 20W recovery work for 2 minutes after a maximum work of 130 W. RR interval data set for a recovery period of minutes. Heart rate recovery is characterized by a quasi-stationary heart rate of 14 beats / minute. FIG. 26 shows the formation of a hysteresis loop for the same patient using QT and RR data. The upper and lower panels show raw data and smoothed data, respectively.

  In both experiments, a sudden stop in motion (FIGS. 21-22 and 25) and a recovery work of 2 minutes after a sudden stop in motion (FIGS. 23-24 and 26) resulted in a quasi-steady state. A sufficiently high heart rate was generated to generate a condition. The maximum work amount in FIG. 26 is four times larger than that in FIG. 25, and the exercise time is about three times longer. For this reason, the work amount of the total exercise performed by the patient who is not coronary artery disease of FIG. 26 is 10 times or more larger than the exercise performed by the patient of the coronary artery disease of FIG. Despite the differences in work, the area values of the QT / RR hysteresis loops in both FIGS. 25 and 26 are approximately the same. This indicates that the patient of FIG. 25 has a much higher degree of ischemia than the patient of FIG. 26 and is a patient with typical coronary artery disease.

Based on the experiment described above and the experiment of Example 9, the definition for a “quasi-stationary” motion (or stimulation) protocol can be defined quantitatively. That is, the quasi-stationary exercise (or stimulation) protocol is a series of two successive stages that increase and decrease heart rate or stimulation (the duration of each stage is 3, 5, 8, or 10 minutes or more) as follows: Refers to that.
1. The duration of each stage is the average peak load heart rate value (about 120 or 130 to 150 or 160 beats / minute) and the average resting heart rate value (about 50 or 60 to 70 or 80 beats / minute). ) And approximately an order of magnitude (eg, at least about 2, 3, 5, 8, or 10 times) longer than the average duration (about 1 minute) to adjust the heart rate while suddenly stopping exercise .
2. The root mean square standard deviation of the first QT or RR interval data set from a smooth, monotonous (for each stage) fit is the peak and rest One order of magnitude (eg, at least about 2, 3, 5, 10 times) less than the average difference between the values of the QT / RR interval.
As shown previously (FIG. 10), the gradual quasi-stationary protocol itself allows significant time-dependent fluctuations to be significantly removed from the measured QT or RR interval data set. The reason is that the duration of these fluctuations is short and the amplitude is small. By fitting the respective QT or RR interval data sets corresponding to each stage to a monotonic function of time, their effects can be further reduced. As a result, the fitted QT interval value between each motion stage can be shown as a nearly monotonic and smooth function of the quasi-stationary changing RR interval value. This function, shown on the (RR spacing, QT spacing) plane, is very well suited to the hysteresis loop shown in FIG. Create a similar loop. Similar to the general hysteresis loop, this loop can be thought of as primarily showing the electrical conduction characteristics of the myocardium.

  It is well known that exercise-induced ischemia changes the state of electrical conduction in the heart. If a particular individual has an ischemic event due to exercise, the combined dispersion-recovery curves obtained from the two experiments corresponding to the ascending and descending phases of the quasi-stationary protocol are different and have a unique quasi-stationary Forming a hysteresis loop can be expected. According to a quasi-stationary protocol, the hysteresis loop is a characteristic of the transient because the evolution of the mean value of the QT and RR intervals occurs very slowly compared to the speed of the transient due to sympathetic / parasympathetic and hormonal control. Hardly depends on. In that case, such hysteresis can provide excellent measurements for changes in the heart's electrical conduction that depend on slow ischemic movements, and generally the health of the heart itself and the health of the cardiovascular system. Can be reflected.

  It should be emphasized in particular that Sarma et al. And N. A. Krahn et al. Do not report studies based on collection of quasi-stationary dependencies similar to those of QT and RR intervals. There is. The reason is that both studies were actually designed for different purposes. On the other hand, they were intended to be based on a non-stationary exercise protocol involving a sudden movement stop at or near the peak of exercise load. These protocols generated different types of QT / RR interval hysteresis loops that were highly indicative of non-stationary sympathetic adrenal transients (see FIG. 9 above). Thus, these prior art embodiments did not include or could not include data characterizing gradual changes in the distribution and recovery characteristics of the heart's electrical conduction, thus preventing exercise-induced ischemia. It did not include any important indications that could be attributed.

[Examples 10 to 12]
[Data Processing in Steps 44 to 45a, b (FIGS. 4A and 4B)]
The following examples illustrate various specific embodiments for performing the process shown in FIG. 4B. Here, the software executed in steps 44B and 45B performs the following main steps.
(I) A sufficiently smooth time-dependent QT and RR by averaging / filtering the data and fitting these averaged data by an exponential function or any other suitable function. Generate a data set.
(Ii) Combine RR and QT data set points corresponding to the same time in pairs, so that the heart rate is on the (T _RR , T _QT ) plane or similar plane in computer memory or its image. Generate smooth QT / RR curves for the rising and falling stages (branches).
(Iii) Close the ends of the QT / RR curves, convert them to a closed loop, determine the area S or a similar size of the region confined by the loop, and subject based on the size of this region Calculate an index that provides quantitative characteristics of cardiovascular health.
Each of these three major steps is comprised of several sub-steps, as shown in FIGS. 11, 13 and 15, 16 and in each method described in more detail later.

ここで、Ｎは対応する（ＲＲまたはＱＴ）未加工データセット内のデーターポイントの数である。このため、ＲＲおよびＱＴ間隔の持続時間および対応するサンプリング時間のポイントは、それらの間の１対１の対応を維持するために同様に平均化される。この方法は、未加工データーの中に存在する高周波ノイズを取り除く。この予備平滑化は周波数領域の中でも説明され、別の方法では、適当なローパスフィルタ処理により実現することができる。この実施例に関係する全ての以下の処理は、平均化されたデーターポイント＜ｕ_i＞₂に対して行われる。また表記を容易にするために角ブラケットを省略する。 [Example 11]
<Method for optimal integration of moving average, exponential function and polynomial fitting>
{1. Averaging / filtering raw data (box 1 in FIG. 11)}
The raw data set consists of two subsets {t ⁱ _RR , T ⁱ _RR }, i = 1, 2,. . , N _RR −1 and {t ⁱ _QT , T ⁱ _QT }, i = 1, 2,. . , N _QT −1. Here, t ⁱ _x and T ⁱ _x indicate the i-th sampling time and the duration of each RR or QT interval (subscript x represents RR or QT). To simplify the notation, subscripts are omitted when subscripts RR and QT can be applied to both subsamples. It is convenient to show each data point as a two-component vector u _i = (t ⁱ , T ⁱ ). The filtering method in this embodiment consists of a moving average of adjacent data points. The moving average for a set of adjacent points is displayed with square brackets indicating the number of points whose subscripts are contained within the averaging operator. Thus, the preliminary data filtering in this substep is i = 1, 2,. . , N−1 is described by the following equation.

Where N is the number of data points in the corresponding (RR or QT) raw data set. Thus, the duration of the RR and QT intervals and the corresponding sampling time points are similarly averaged to maintain a one-to-one correspondence between them. This method removes high frequency noise present in the raw data. This pre-smoothing is also explained in the frequency domain, and in another way it can be realized by suitable low-pass filtering. All the following processing relating to this embodiment is performed on the averaged data point <u _i > ₂ . Also, square brackets are omitted for ease of notation.

の予備的評価から構成する。このアルゴリズムはデーターセットを順次ソーティングして、ＲＲまたはＱＴ間隔のＭ個の最小値に対応するＭ個のデーターポイントを選択し、次に結果を平均する。（この実施例では、Ｍ＝１０である。）第１に、このアルゴリズムによって、最短の間隔

に相当するｕ−ベクトル

が明らかとなる。次に、このデーターポイントはデーターセット｛ｕ_i｝から除かれ、アルゴリズムは残りデータセットをソーティングして、最短の間隔

に相当するｕ−ベクトル

が再度明らかとなる。最小ポイントが再びデーターセットから除かれ、最短の間隔

に相当するＭ番目のｕ−ベクトル

が明らかとなるまで、ソーティングは何度も繰り返される。次に、アルゴリズムは全てのＭ個の対を平均し、下記の式として定義された平均最小時間座標

を計算する。

最後に、このアルゴリズムによって、

に最も近いサンプリング時定数

が決定される。このようにして、ＲＲおよびＱＴデータセットに対して、それぞれ

に到達する。 {2. Preliminary evaluation of minimum t-coordinate (box 2 in FIG. 11)}
This substep is the minimum time constant of the first averaged RR interval and QT interval data sets, respectively.

Consists of preliminary evaluations. The algorithm sorts the data set sequentially, selects M data points corresponding to the M minimum values of the RR or QT interval, and then averages the results. (In this example, M = 10.) First, the algorithm provides the shortest spacing.

U-vector corresponding to

Becomes clear. This data point is then removed from the data set {u _i } and the algorithm sorts the remaining data set to the shortest interval

U-vector corresponding to

Becomes clear again. The minimum point is removed from the data set again and the shortest interval

Mth u-vector corresponding to

Sorting is repeated many times until it becomes clear. The algorithm then averages all M pairs, and the average minimum time coordinate defined as

Calculate

Finally, this algorithm

Sampling time constant closest to

Is determined. Thus, for RR and QT data sets respectively

To reach.

に対する第１の補正を求める。アルゴリズムのこの部分は、下記の式の関数により最初にフィルタ処理されたデーターセット｛ｕ_i｝＝｛ｔⁱ，Ｔⁱ｝のＴ成分の反復する指数関数のフィッティングに基づく。
Ｔ（ｔ）＝Ａｅｘｐ［β｜ｔ−ｔ^m｜］ （式１０．３）
ここで、ｔ^mは、前述のサブステップ（すなわち、ＲＲ間隔に対しては

、ＱＴ間隔に対しては

）において決定された対応する最小値の時刻である。フィッティングは、両方のブランチに対して定数Ａおよびｔｍの値は同じで定数βの値は異なるｕ（ｔ）の下降（ｔ＜ｔ^m）および上昇（ｔ＞ｔ^m）ブランチに対して別々に行われる。Ａの初期値は、サブステップ２における予備評価、すなわち、

（すなわち、ＲＲ間隔に対しては

およびＱＴ間隔に対しては

）から得られる。各ブランチおよび各データーのサブセットに対する定数βの初期値は、対応するブランチ全体の式（１０．３）によるフィッティングからの初期の二乗平均平方根の偏差σ＝σ₀が最小であるという要求事項から得られる。その後の各繰返しサイクルの中で、定数Ａおよびβの新しい値が各ブランチに対して決定される。各繰返しサイクルの開始時に、定数Ａが前のステップから取得され、βの値の数値が調整されて、偏差値σを最小にする。そのような反復は、平均二乗偏差σが小さくなっていく間は繰り返され、σが最小値σ＝σ_minに達すると停止される。このサブステップの終了時に、このアルゴリズムは補正値Ａおよびβ並びにσ^RR _minおよびσ^QT _minのそれぞれの値を出力する。 {3. First correction of preliminary minimum coordinate (box 3 in FIG. 11)}
In this substep, the minimum coordinate

A first correction for is obtained. This part of the algorithm is based on an iterative exponential fitting of the T component of the data set {u _i } = {t ⁱ , T ⁱ }, which was first filtered by the function of the following equation:
T (t) = Aexp [β | t−t ^m |] (Formula 10.3)
Where t ^m is the aforementioned sub-step (ie, for the RR interval)

For QT interval

) Is the corresponding minimum value time determined in (1). The fitting is done separately for the falling (t <t ^m ) and rising (t> t ^m ) branches of u (t) with the same values of constants A and tm but different values of constant β for both branches. Done. The initial value of A is a preliminary evaluation in sub-step 2, ie,

(Ie, for RR interval

And for QT interval

) The initial value of the constant β for each branch and subset of data is obtained from the requirement that the initial root mean square deviation σ = σ ₀ from the fitting according to equation (10.3) for the corresponding entire branch is minimal. It is done. During each subsequent iteration cycle, new values of constants A and β are determined for each branch. At the beginning of each iteration cycle, the constant A is obtained from the previous step and the numerical value of β is adjusted to minimize the deviation value σ. Such iteration is repeated while the mean square deviation σ decreases, and is stopped when σ reaches the minimum value σ = σ _min . At the end of this substep, the algorithm outputs the correction values A and β and the values of σ ^RR _min and σ ^QT _min , respectively.

量ｐを平均化窓の幅と呼ぶ。計算はｐの様々な値に対して実行され、次にサブステップ３で決定されたパラメータＡ、ｔ^mおよびβの値を用いて、式（１０．３）によるフィッティングからセット｛ｕ_i ^p｝＝｛ｔ_i ^p，Ｔ_i ^p｝の平均二乗偏差を最小にすることによって、平均化窓の幅の最適値ｐ＝ｍが求められる。データーセット（ＲＲおよびＱＴ）の各成分に対してこの方法を実行した後、予備的に平滑化されたデーターセット｛ｕ_i ^m｝＝｛ｔ_i ^m，Ｔ_i ^m｝に到達する。ここでは、以下の式が成立する。

このようなデーターセットにおけるポイント数は、Ｎ_m＝Ｎ−ｍ＋１である。ここで、Ｎはサブステップ１で初期のフィルタリングを行った後のデーターポイントの数である。 {4. Pre-smoothing curve (box 4 in FIG. 11)}
This sub-step consists of calculating a series of moving averages over p consecutive points for each data subset {u} as follows:

The quantity p is called the width of the averaging window. The calculation is performed on various values of p, then using the values of parameters A, t ^m and β determined in sub-step 3, set from fitting according to equation (10.3) {u _i ^p } By minimizing the mean square deviation of = {t _i ^p , T _i ^p }, the optimum value p = m of the width of the averaging window is obtained. After performing this method for each component of the data set (RR and QT), a pre-smoothed data set {u _i ^m } = {t _i ^m , T _i ^m } is reached. Here, the following formula is established.

The number of points in such a data set is N _m = N−m + 1. Here, N is the number of data points after the initial filtering in substep 1.

との間に位置するように、最初に全てのデーターポイント｛ｔⁱ，Ｔⁱ｝を規定する。これらのデーターポイントは、下記のセットから上付き文字の値ｉを得る。

｛ｔ_i｝（ｉ∈Ｉ）の間で最も早い時刻および最も遅い時刻に対応する下付き文字の値を、ｉ₀∈Ｉおよびｉ_q∈Ｉで表示する。従って、ｉ₀＝ｍｉｎ｛Ｉ｝およびｉ_q＝ｍａｘ｛Ｉ｝と設定する。ポイントの数をＲＲデーターについてはｑ_RRで、ＱＴデーターについてはｑ_QTによりそのようなセットの中で表示して、平均化窓の幅ｑを２つの最小値ｑ＝ｍｉｎ｛ｑ_QT，ｑ_RR｝として求める。ここで、最小値の近くのデーターポイントに対する移動平均を下記のように記載する。

そのような修正された平均化は、下付き文字ｉがｉ₀からｉ_qにわたる全ての連続したデーターポイントｕ_iに適用される。最終的な（平滑化された）データーセットは、ｉ＝１，２，．．，ｉ₀−１に対して式（１０．４）により定義された最初のｉ₀−１個のデーターポイント、ｉ＝ｉ₀，ｉ₀＋１，．．，ｉ_qに対して式（１０．７）により定義されたｑ個のデーターポイント、およびｉ＝ｉ_q＋１，ｉ_q＋２，．．．，Ｎ−ｍ＋１に対して再度式（１０．４）により定義されたＮ−ｉ_q−ｍ個のポイントから構成する。この最終的なセット

は、下記の式として示すことができる。

サブステップ１０．５の最後において、このアルゴリズムはＱＴおよびＲＲ間隔の最終的な最小値および対応する時刻を決定する。アルゴリズムは、平滑化されたデーターセット（１０．８）をソートし、

の最小値に相当する

を決定する。これは、下記の式として記載することができる。

{5. Correction of pre-smoothing curve and second correction of minimum values of QT and RR and their coordinates (box 5 in FIG. 11)}
In this sub-step, we redefine the moving average near the minimum of the quantity T (RR or QT interval) (using a smaller averaging window). This is useful to avoid distortion of the sought (fitting) curve T = T (t) near its minimum value. This algorithm has T ⁱ

First, all the data points {t ⁱ , T ⁱ } are defined so as to be between the two. These data points get the superscript value i from the following set:

{T _i} the value of subscript corresponding to the earliest time and latest time between (i∈I), displayed in i ₀ ∈I and i _q ∈I. Therefore, i ₀ = min {I} and i _q = max {I} are set. The number of points is indicated in such a set by q _RR for RR data and q _QT for QT data, and the averaging window width q is expressed as two minimum values q = min {q _QT , q _RR }. Here, the moving average for the data points near the minimum value is described as follows.

Such a modified averaging is applied to all consecutive data points u _i with subscript i ranging from i ₀ to i _q . The final (smoothed) data set is i = 1, 2,. . , I ₀ −1, the first i ₀ −1 data points defined by equation (10.4), i = i ₀ , i ₀ +1,. . , I _q , _q data points defined by equation (10.7), and i = i _q +1, i _q +2,. . . , N−m + 1, again, N−i _q −m points defined by the equation (10.4). This final set

Can be expressed as:

At the end of substep 10.5, the algorithm determines the final minimum value of QT and RR intervals and the corresponding time. The algorithm sorts the smoothed data set (10.8)

Corresponds to the minimum value of

To decide. This can be described as:

は、それぞれ、下降

および上昇

ブランチに対応する２つのサブセット｛ｕⁱ _-｝および｛ｕⁱ ₊｝に分割される。時間の原点を最小ポイント

に移動し、下降ブランチの軸方向を変更する。これにより、これらの再定義された変数は、下記のように与えられる。

最小ポイントが両方のブランチの中に含まれることに注意されたい。次に、各ブランチを下記の形式の４次多項式関数を用いてフィッティングすることによって、直線回帰を実行する。
Ｔ±（ｔ±）＝ａ±ｔ±⁴＋ｂ±ｔ±³＋ｃ±ｙ±²＋ｄ （式１０．１１）
この関数はｔ±＝０において最小値を有する必要があるため、直線項はこの方程式には含まれない。ｄの値は、下記の式で定義される。

計算の都合上、変数ｕ±はｚ±＝Ｔ±−ｄに変換され、次に係数ａ±、ｂ±およびｃ±が、下記のエラーの方程式を最小にする条件から決定される。

この式では、総和が対応するブランチの全てのポイントにわたって実行される。このように、代表的な直線方程式の４つの同様なセット、すなわち２つのＲＲおよびＱＴデータセットの各ｕ±ブランチに対して２つ、が得られる。これらの方程式の形式は、下記のようになる。

全ての総和が、ブランチの全てのポイントにわたって実行される。ここで、ＱＴおよびＲＲ間隔の時間依存曲線を、下記のように示すことができる。
Ｔ_RR（ｔ）＝Ｔ_RR ^-（ｔ−ｔ_min ^RR），ｔ≦ｔ_min ^RR （式１０．１５）
＝Ｔ_RR ⁺（ｔ−ｔ_min ^RR），ｔ≧ｔ_min ^RR
および
Ｔ_QT（ｔ）＝Ｔ_QT ^-（ｔ−ｔ_min ^QT），ｔ≦ｔ_min ^QT （式１０．１６）
＝Ｔ_QT ⁺（ｔ−ｔ_min ^QT），ｔ≧ｔ_min ^QT
ここで、Ｔ_RR±（ｔ）およびＴ_QT±（ｔ）は、式（１０．１１）によって与えられた４次の多項式であり、ｔ_min ^RRおよびｔ_min ^QTは式（１０．９）によって定義された対応する最小値の時間座標である。このように、式（１０．１０）〜式（１０．１６）は、式（１０．９）により定義された最小値を用いて、最終的な平滑なＱＴおよびＲＲ曲線を決定する。 {6. Final fitting and final smooth QT and RR curves (box 6 in FIG. 11)}
In this substep, the parameters of the function showing the final smooth QT and RR curves are found. First of all, each data set

Are descending respectively

And rise

It is divided into two subsets {u ⁱ ₋ } and {u ⁱ ₊ } corresponding to the branches. Minimum point with time origin

To change the axial direction of the descending branch. This gives these redefined variables as follows:

Note that the minimum point is included in both branches. Next, linear regression is performed by fitting each branch with a quartic polynomial function of the form:
T ± (t ±) = a ± t ± ⁴ + b ± t ± ³ + c ± y ± ² + d (Formula 10.11)
Since this function needs to have a minimum at t ± = 0, the linear term is not included in this equation. The value of d is defined by the following equation.

For computational convenience, the variable u ± is converted to z ± = T ± −d, and then the coefficients a ±, b ±, and c ± are determined from the conditions that minimize the error equation below.

In this equation, the summation is performed over all points of the corresponding branch. In this way, four similar sets of representative linear equations are obtained, two for each u ± branch of the two RR and QT data sets. The form of these equations is as follows:

All sums are performed across all points in the branch. Here, the time-dependent curves of the QT and RR intervals can be shown as follows.
T _RR (t) = T _RR ⁻ (t−t _min ^RR ), t ≦ t _min ^RR (Formula 10.15)
= T _RR ⁺ (t−t _min ^RR ), t ≧ t _min ^RR
And T _QT (t) = T _QT ⁻ (t−t _min ^QT ), t ≦ t _min ^QT (Formula 10.16)
= T _QT ⁺ (t−t _min ^QT ), t ≧ t _min ^QT
Here, T _RR ± (t) and T _QT ± (t) are quartic polynomials given by equation (10.11), and t _min ^RR and t _min ^QT are obtained by equation (10.9). The corresponding minimum time coordinate defined. Thus, equations (10.10) to (10.16) determine the final smooth QT and RR curves using the minimum value defined by equation (10.9).

ここで、Ωは閉じたヒステリシスループによって形成された境界を有する（Ｔ_RR，Ｔ_QT）平面上の領域であり、ρ（Ｔ_RR，Ｔ_QT）は、負でない（重み）関数である。この実施例では、Ｓが領域Ωの面積と一致するようにρ（Ｔ_RR，Ｔ_QT）＝１としたり、またはＳが（ｆ，Ｔ_QT）平面上のヒステリシスループ内部の領域の微小な面積と一致するように、ρ＝１／（Ｔ_RR）²と設定することができる。ここで、ｆ＝１／Ｔ_RRは心拍数である。 {7. Final smoothing hysteresis loop (box 7 in FIG. 11)}
In this substep, the software first generates a dense (N + 1) point time grid τ _k = t _start + k (t _end −t _start ) / N. Here, k = 0, 1, 2,. . , N (N = 1000 in this example), t _start and t _end are actual time values at the _start and _end of the measurement, respectively. This sub-step then calculates the values of the four functions T _RR ± (τ−t _min ^RR ) and T _QT ± (τ−t _min ^QT ) on the grid. This corresponds to the final smoothing curves (T _RR ± (τ−t _min ^RR ), T _QT ± (τ−t _min ^QT ) in the computer's memory for the rising (+) and falling (−) branches, respectively. )) Is displayed as a parameter. This method is computationally equivalent to the analytical removal of time. The software then plots these smoothing curves on the (T _RR , T _QT ) plane or another similar plane such as (f _RR , T _QT ). Here, f = 1 / T _RR is an instantaneous heart rate. Finally, the algorithm adds a set of points to the curve that connect the end of the lower (falling) branch to the start of the upper (rising) branch, thereby creating a closed QT / RR hysteresis loop. .
{8. Measurement of the area enclosed by the QT / RR hysteresis loop (box 8 in FIG. 11)} In this substep, the measurement of the area inside the QT / RR hysteresis loop is calculated by numerically evaluating the following integral: (See definition above).

Here, Ω is a region on the (T _RR , T _QT ) plane having a boundary formed by a closed hysteresis loop, and ρ (T _RR , T _QT ) is a non-negative (weight) function. In this embodiment, ρ (T _RR , T _QT ) = 1 is set so that S matches the area of the region Ω, or S is a small area of the region inside the hysteresis loop on the (f, T _QT ) plane. So that ρ = 1 / (T _RR ) ² . Here, f = 1 / T _RR is the heart rate.

窓の幅ｍｆは数の上で変化して、フィッティングの最適な平滑性および精度を実現する。最適に平均化されたデーターポイントは、対応する平面上に平滑な曲線を形成する（図１２、図１４）。 [Example 12]
<Sequential moving average method>
{1. Raw data averaging / filtering}
This substep (similar to 10.1) consists of the averaging (filtering) of the raw data according to equation (10.4). This is performed for a preliminary set of values of the averaging window width p.
{2. Smoothing with final moving average}
This sub-step includes the subsequent final smoothing of the pre-smoothed data indicated by the set {u _i ^p } determined in the previous step as described below.

The window width mf varies in number to achieve optimal smoothness and accuracy of the fitting. The optimally averaged data points form a smooth curve on the corresponding plane (FIGS. 12 and 14).

{3. Measurement of the area inside the QT / RR hysteresis loop}
The method of this sub-step is the same as that defined in Step 8 of Example 10 above.

[Example 13]
<Optimization nonlinear transformation method>
FIG. 15 illustrates the main steps in the data processing procedure including this nonlinear transformation method. The first three stages for the RR and QT datasets are very similar, each resulting in a fitted bird-shaped curve T ^RR = T ^RR (t) and T ^QT = T ^QT (on a dense time grid. t) is calculated. When both bird-shaped curves are calculated, the data processing procedure is actually completed. The reason for this is that after proper synchronization, these two dependencies, along with parametrically closed lines, show the sought hysteresis loop in the (T ^RR , T ^QT ) plane or similar plane. The inventors describe these methods in general terms that are equally applicable to QT and RR interval data sets, and indicate specific moments when there are differences in the algorithms.

{1. Preliminary data processing stage}
[T _k }, k = 1, 2,. . , N is the set of measured RR or QT interval durations and {t _k } is the corresponding set of times, t ₁ and t _N are the start time and end time (section) of the entire recording. The first three similar steps (steps 1 to 3 in FIG. 15) are a preliminary step, a second nonlinear transformation step, and a calculation step. A more detailed data flow chart for these stages is shown in FIG. The preliminary stages shown in Box 1 to Box 7 of FIG. 16 are combinations of conventional data processing methods and include the following processes: (1) smoothing (averaging) the data set; (2) Determination of the area near the minimum value, (3) Fitting a secondary parabola to the data in this area, (4) Checking the consistency of the results, (5) Renormalization and centering of the minimum value data, (6) Motion It includes processing of separation of data sections outside the region, separation of rising and falling branches, and finally (7) filtering of data sections near the minimum value. These steps will be described in more detail.

ｍが一定であり不明確性が生じる恐れがない場合は、下付き文字ｍは以下では省略される。このアルゴリズムの次のステップは、放物線のフィッティングが行われる時間間隔（放物線フィッティング間隔）の初期決定である。このステップを、図１６のボックス２の中で示す。一定の条件が満足されない場合は、このデーターのサブセットを後の段階で再定義できることに注意する。このアルゴリズムを現在実現するために、この領域を｛ｔ_i，Ｔ_i ^RR｝および｛ｔ_i，Ｔ_i ^QT｝のデーターセットに対して別々に定義した。データーセット｛ｔ_i，Ｔ_i ^RR｝に対して、初期の放物線のフィッティング区間が、ｉ₁とｉ₂との間のｉの全ての連続した値を有するデーター区間｛ｔ_i，Ｔ_i ^RR｝として定義される。ここで、ｉ₁＝ｎ_min−△ｍおよびｉ₂＝ｎ_min＋△ｍであり、整数のパラメータｎ_minおよび△ｍは次のように定義される。数値ｎ_minは、平均化されたデーターセット｛＜ｔ_i＞，＜Ｔ_i＞｝上の最小値に最も近い時刻の本来のセットの中の時刻

を決定する。換言すると、

は、平均のＲＲ間隔＜Ｔ_i ^RR＞の最小値に対応する平均時刻＜ｔ_M＞、すなわち、下記の条件により定義された下付き文字の値Ｍを有する時刻に最も近い。

△ｍの値は、条件△ｍ＝２ｍによってｍの値にリンクされる。△ｎとｍとの間のリンクは、アルゴリズムが安定で一貫しているという要求事項から生ずる。この一貫性は、移動平均および二次フィッティングによって得られた曲線の最小値の位置はほぼ同じであったという要求事項である。他方、二次フィッティングは最小値の直ぐ近くの局所的なデーターのみを説明する目的で行われるため、ｍの値は十分に小さいことも重要である。 Box 1 of FIG. 16 shows the moving average and itself includes two steps. First, the initial value of the parameter m of the moving average procedure is specified by the following equation.
m = max {γ (N / N _c ), _mmin } (Formula 12.1)
Here, r (x) is an integer closest to x, the values of N _c and m _min are, based on the amount of random fluctuations in the number N and data of data points (the size of the random component of the data) Selected. In this example, N _c = 100 and _mmin = 3. The value of m can be redefined repeatedly at a later stage, the selection of which will be explained at that time. Next, a moving average is calculated for the {t _i } and {T _i } data sets using a predetermined averaging parameter m as follows:

If m is constant and there is no risk of ambiguity, the subscript m is omitted below. The next step in this algorithm is the initial determination of the time interval (parabolic fitting interval) during which the parabola fitting is performed. This step is shown in box 2 of FIG. Note that this subset of data can be redefined at a later stage if certain conditions are not met. In order to currently implement this algorithm, this region was defined separately for the {t _i , T _i ^RR } and {t _i , T _i ^QT } data sets. For a data set {t _i , T _i ^RR }, a data interval {t _i , T _i ^RR } in which the initial parabolic fitting interval has all consecutive values of i between i ₁ and i ₂ . Is defined as Here, i ₁ = n _min −Δm and i ₂ = n _min + Δm, and the integer parameters n _min and Δm are defined as follows. The number n _min is the time in the original set of times closest to the minimum on the averaged data set {<t _i >, <T _i >}

To decide. In other words,

Is closest to the average time <t _M > corresponding to the minimum value of the average RR interval <T _i ^RR >, that is, the time having the subscript value M defined by the following conditions.

The value of Δm is linked to the value of m by the condition Δm = 2m. The link between Δn and m arises from the requirement that the algorithm be stable and consistent. This consistency is a requirement that the position of the minimum of the curve obtained by moving average and quadratic fitting was approximately the same. On the other hand, it is also important that the value of m is sufficiently small since secondary fitting is done for the purpose of explaining only local data in the immediate vicinity of the minimum value.

ここで、Ｒは、二次放物線を用いてフィッティングされるデーターの部分を決定するパラメータ０＜Ｒ＜１である。この計算では、Ｒ＝１／８＝０．１２５と設定して、間隔の持続時間の底部１２．５％に相当するデーターポイントを放物線とフィッティングする。このため、下付き文字ｉ₁は、条件（式１２．４）が満足されるようなｉの第１の（最小の）値である。同様に、ｉ₂は、条件（式１２．４）が満足されるようなｉの最後の（最大の）値である。次に、初期のフィッティング領域が、以下の非平均化データーのサブセット｛（ｔ_i，Ｔ_i ^QT）：ｉ＝ｉ₁，ｉ₁＋１，ｉ₁＋２，．．，ｉ₂｝として定義される。この方法は、ＲＲデーターセットの二次多項式フィッティングに対して初期のデーター区間を決定するために用いることもできる。 For the data set {t _i , T _i ^QT }, the initial parabolic fitting interval is defined as the subset of data consisting of all consecutive points belonging to the bottom of the non-averaged {T _i ^QT } data set Is done. We the i ₁ as this is defined as a first (minimum) numerical i as follows.

Here, R is a parameter 0 <R <1 that determines a portion of data to be fitted using a secondary parabola. In this calculation, R = 1/8 = 0.125 is set, and a data point corresponding to the bottom 12.5% of the interval duration is fitted with a parabola. Thus, the subscript i ₁ is the first (minimum) value of i that satisfies the condition (Equation 12.4). Similarly, i ₂ is the last (maximum) value of i such that the condition (Equation 12.4) is satisfied. Next, the initial fitting region is the following subset of non-averaged data {(t _i , T _i ^QT ): i = i ₁ , i ₁ +1, i ₁ +2,. . , I ₂ }. This method can also be used to determine the initial data interval for quadratic polynomial fitting of the RR data set.

これは、このデーターのサブセットに対して通常の直線回帰を用いて行われる。次のステップ（図１６のボックス４）では、放物線が最小値を有するかどうか（すなわち、Ｐ₁＞０かどうか）をチェックする。この条件が満足される場合、放物線のフィッティング間隔に対する決定および放物線自身をフィッティングすることが完了する。満足されない場合は、図１６のボックス１ａ〜ボックス３ａで示すループに入る。第１のステップは、ＲＲ間隔データセットに対して用いられた上記の第２のステップに似ており、放物線のフィッティングに対して拡張された区間を定義する。このことは、図１６のボックス１ａに示すように式（１２．２）に従って、ｍをｍ＋１に置き換え、この新しいｍの値を用いて△ｍ＝２ｍを再定義し、新しい平均化されたデーターセットを計算することによって行われる。これにより、ｎ_minの新しい値および並びにｉ₁＝ｎ_min−△ｍおよびｉ₂＝ｎ_min＋△ｍの新しい値をｍの新しい値で評価することができる（ボックス２ａ）。次に、放物線のフィッティングが再度行われ（ボックス３ａ）、放物線が最小値を有する条件Ｐ₁＞０が再びチェックされる。その条件が満足される場合、フィッティングの間隔および二次フィッティング係数を決定する手順が完了する。満足されない場合は、ｍをｍ＋１で置き換えて、条件Ｐ₁＞０が満足されるまで何回も処理を繰り返す。この条件は、二次放物線の極値が実際に最小値であることを保証する。まさに運動プロトコルの設計により、平均の心拍数が負荷段階内のどこかで一定の最大値に到達し、これにより、対応する平均のＲＲ間隔が最小になる。このことは、そのように定義されたデーター区間が存在し１つしかないこと、このため、二次放物線の係数がうまく定義されていることを保証する。要するに、発明者らは平均化されたデーターセットの最小値の周りに集中し、最小値（Ｐ₁＞０）を有する二次放物線を生成する最も短いデーター区間を用いる。 In the next step, indicated by box 3 in FIG. 12.2, the data in this region is fitted (for each data set) by a parabola, as the data is approximately represented by the following equation:

This is done using normal linear regression on this subset of data. The next step (box 4 in FIG. 16) checks whether the parabola has a minimum value (ie, whether P ₁ > 0). If this condition is satisfied, the determination for the parabola fitting interval and the fitting of the parabola itself is complete. If not satisfied, the loop shown by box 1a to box 3a in FIG. 16 is entered. The first step is similar to the second step used above for the RR interval data set and defines an extended interval for parabolic fitting. This replaces m with m + 1 according to equation (12.2) as shown in box 1a of FIG. 16, redefines Δm = 2m using this new value of m, This is done by calculating the set. This allows the new value of n _{min and} the new value of i ₁ = n _min −Δm and i ₂ = n _min + Δm to be evaluated with the new value of m (box 2a). Next, parabola fitting is performed again (box 3a) and the condition P ₁ > 0 with the parabola having the minimum value is checked again. If the condition is satisfied, the procedure for determining the fitting interval and the secondary fitting coefficient is completed. If not satisfied, m is replaced with m + 1, and the process is repeated many times until the condition P ₁ > 0 is satisfied. This condition ensures that the extreme value of the secondary parabola is actually the minimum value. Exactly due to the design of the exercise protocol, the average heart rate reaches a certain maximum value somewhere in the load phase, which minimizes the corresponding average RR interval. This ensures that there is only one data interval so defined and, therefore, that the coefficients of the secondary parabola are well defined. In short, we use the shortest data interval that concentrates around the minimum of the averaged data set and produces a secondary parabola with the minimum (P ₁ > 0).

Parabola fitting defines two important parameters of the data processing procedure: the minimum position (t _min , T _min ) on the (t, T) plane as follows.
t _min = −P ₂ / 2P ₁ , T _min = P ₃ −P ₂ ² / 4P ₁ (Formula 12.6)
These parameters are final in the sense that the final fitting curve of the present application always passes and has the minimum value at the point (t _min , T _min ). The coordinates (t _min , T _min ) thus constitute the final bird-like fitting curve parameters. When the minimum value ordinate T _min is found, the T data set can be transferred as follows.
y _k = T _k / T _min (Formula 12.7)
The inventors also regard the abscissa of the parabolic minimum value tmin as the time origin, and define the time component of the data point as follows.
t _i ⁻ = t _i −t _min , t _i ≦ 0 (formula 12.8)
t _j ⁺ = t _j −t _min , t _j > 0
These two data transformations are shown in box 5 of FIG. We also limit the data set to only the duration of exercise plus some short back and forth intervals (box 5). The conditions t _i ⁻ ≦ 0 and t _j ⁺ > 0 define the rising {t _i ⁻ , T _i } and falling {t _j ⁺ , T _j } branches, respectively, so that the corresponding data points are easily identified and separated. (Box 6). When lowering and rising of the load stage of duration and a t _d and t _a, respectively, the points on the descending branch t _i ^- by separating at <-t _d, the point on the rising branch t _j ^+> at t _a , The original set can be reduced (box 6 in FIG. 16). Thereby, the minimum value i ₀ of the subscript i for the descending branch and the maximum value j _max of the subscript j on the ascending branch are determined.

The final stage of the preliminary data processing is conditional sorting (box 7). This conditional sorting moves all consecutive points so that at least one of them is below the parabolic minimum. It is necessary to move the point below the minimum value because a non-linear transformation is possible only when y _i ≧ 1. By removing only the distant points below the minimum value y = 1, a bias is created in the data. For this reason, it is necessary to remove all sections of data near the minimum value. However, it should be recalled that these points have already been taken into account in the above-mentioned secondary filtering process that determines (t _min , T _min ) according to equation (12.6).

Thus, preliminary data processing results in two data sets corresponding to the steps of lowering and increasing the load. The descending data set consists of consecutive pairs (t _i ⁻ , y _i ) ≡ (t _i ⁻ , y _i ⁻ ), i = i ₀ , i ₀ +1, i ₀ +2,. . , I _max . Here, i _max is determined by conditional sorting as the maximum subscript i value that still satisfies the condition y _i ≧ 1. Similarly, the ascending data set consists of consecutive pairs (t _i ⁺ , y _i ) ≡ (t _i ⁺ , y _i ⁺ ), j = j _min , j ₀ +1, j ₀ +2,. . , J _max . Here, j _min is determined by conditional sorting as the first subscript j value on the rising branch starting from a position where the condition y _j ≧ 1 is satisfied for all subsequent j.

第２の場合γ＝２は、下記の式により説明される。

両方の関数は、ｕ＝１で最小値ｙ＝１になる。連続パラメータγ＞０に依存するそのような関数の実施例は、下記により与えられる。
ｆγ（ｕ）＝Ａγ（１／（γ²＋ｂγｕ）＋γ²ｂγｕ／（γ²ｂγｕ＋１）），
ｂγ≡１＋γ＋１／γ （式１２．１７）
パラメータｂγは、ｆγ（ｕ）がｕ＝１で最小値を有する条件から決定され、係数Ａγは条件ｆγ（１）＝１によって決定され、下記の式を生成する。
Ａγ＝（γ³＋γ²＋γ＋１）／（γ³＋γ²＋２γ） （式１２．１８）
下記の数値の例では、発明者らは２つの値１および２を取るγを有する離散パラメータの事例を用いる。元のセットおよび変換されたセットを、図１７（ＲＲ間隔）および図１８（ＱＴ間隔）で示す。両方の図面のパネルＡは、（ｔ，Ｔ）平面上の元データセットを示し、パネルＢおよびパネルＣは、それぞれγ＝１およびγ＝２に対する（ｔ，ｕ）平面上の変換されたセットを示す。パネルＡおよびパネルＣ上の放物線の最小値を、円で囲ったアスタリスクで示す。元のデーターポイントは、単調でない曲線（平均曲線）の近くに集中している。変換された平面上のデーターポイントは、単調に増加する（平均）曲線の周りに集中する。その上、図面は、変換がゼロから最後のゼロでない値までのｔ＝ｔ_minの近傍で、平均曲線の勾配を変化することを示している。（ｔ−ｔ_min，ｕ）平面上では、放物線の最小値に対応するポイントは（０，１）である。このポイントも、図１７および図１８のパネルＢおよびパネルＣ上で、円で囲ったアスタリスクによってマークされる。 {2. Second data processing}
In the second stage, the inventors introduce a radically new method of nonlinear regression with two consecutive optimal nonlinear transformations. The idea of this method introduces two appropriate non-linear transformations that depend on several parameters of the dependent and independent variables y = f (u) and u = φ (t) for each branch, and these Selecting the value of the parameter in such a way that the component f (φ (t _i )) of the transformation provides an approximation to y _i . Let (t _i ⁻ , y _i ⁻ ) and (t _i ⁺ , y _i ⁺ ) be cut, conditional sort, and normal corresponding to the descending and rising branches that show the dynamics of the RR or QT interval during the exercise Data set. The non-linear transformation y⇒u is defined by the following smoothing function.
y = fγ (u) (Formula 12.9)
Since this smoothing function has a unit minimum at u = 1, fγ (1) = 1, fγ ′ (1) = 0, and fγ (u) is monotonous when u ≧ 1. It increases and decreases monotonously when u ≦ 0. The subscript y represents a discrete or continuous set of parameters and indicates a particular selection of such a function. Let fγ ⁻ (u) denote a monotonically decreasing branch of fγ (u), and fγ ⁺ (u) denote a monotonically increasing branch. Then, it can be described as follows.
fγ (u) = fγ ⁻ (u), when u ≦ 1 (Formula 12.10)
^{fγ (u) = fγ + (} u), the case of ^{u ≧ 1 u = gγ + (} y) and u = G.gamma ^- a (y), and inverse function for each branch of fγ (u). Function gγ ⁺ (y) and fγ ⁺ (u) monotonically increasing, whereas the function gγ ^- (y) and fγ ^- (u) is monotonically decreasing. {T _i , y _i ⁻ } denotes the data set decrease interval, ie, data for t _i <t _min , and {t _i , y _i ⁺ } denotes the data set rise interval, ie, t _i > t _min Let's show the data. The following transformation is uγ _{, i} ⁻ = gγ ⁻ (y _i ⁻ ) (Formula 12.11)
The monotonically decreasing (on average) data set {t _i , y _i ⁻ } is mapped to the monotonically increasing data set as:
{T _i , y _i ⁻ } => {t _i , g γ ⁻ (y _i ⁻ )} ≡ {t _i , u γ _{, i} ⁻ } (Formula 12.12)
Moreover, the average slope of the original data set, decreases as t _i approaches t _min, finally disappears at a minimum value. In contrast, the average slope of the transformed data set is not always zero at t = t _min . Similarly, the following transformation is expressed as uγ _{, j} ⁺ = gγ ⁺ (y _j ⁺ ) (Equation 12.13)
The monotonically increasing (on average) data set {t _i , y _i ⁺ } is mapped to the monotonically increasing data set as follows:
{T _j , y _j ⁺ } ⇒ {t _j , g γ ⁺ (y _j ⁺ )} ≡ {t _j , u γ _{, j} ⁺ } (Formula 12.14)
In this embodiment, we use a discrete parameter γ that takes two values γ = 1 and γ = 2 corresponding to two specific choices of nonlinear y-transform. The first case (γ = 1) is described by the following equation.

The second case γ = 2 is described by the following equation.

Both functions have a minimum value y = 1 at u = 1. An example of such a function depending on the continuous parameter γ> 0 is given by
fγ (u) = Aγ (1 / (γ ² + bγu) + γ ² bγu / (γ ² bγu + 1)),
bγ≡1 + γ + 1 / γ (formula 12.17)
The parameter bγ is determined from the condition in which fγ (u) has a minimum value when u = 1, and the coefficient Aγ is determined by the condition fγ (1) = 1 to generate the following equation.
Aγ = (γ ³ + γ ² + γ + 1) / (γ ³ + γ ² + 2γ) (formula 12.18)
In the numerical example below, we use the discrete parameter case with γ taking two values 1 and 2. The original and transformed sets are shown in FIG. 17 (RR interval) and FIG. 18 (QT interval). Panel A in both drawings shows the original data set on the (t, T) plane, and panel B and panel C are transformed sets on the (t, u) plane for γ = 1 and γ = 2, respectively. Indicates. The minimum parabola on panels A and C is indicated by an asterisk enclosed in a circle. The original data points are concentrated near a non-monotonic curve (average curve). Data points on the transformed plane are concentrated around a monotonically increasing (average) curve. Moreover, the drawing shows that the transformation changes the slope of the average curve in the vicinity of t = t _min from zero to the last non-zero value. On the (t−t _min , u) plane, the point corresponding to the minimum value of the parabola is (0, 1). This point is also marked by a circled asterisk on panels B and C of FIGS.

A pair of new time variables τ ^- and τ ⁺ are introduced for the descending and rising branches, respectively. The inventors count them from the parabola fitted and the abscissa of the set minimum as follows.
τ _i ⁻ = t _min −t _i , t _i ≦ t _min , i = 1, 2,. . , I ⁻ (Formula 12.19)
τ _j ⁺ = t _j −t _min , t _j ≧ t _min , j = 1, 2,. . , J ⁺
Where I ⁻ and J ⁺ are the number of data points on the descending and ascending branches, respectively. At the same time, using the time reversal given by the first line of equation (12.19) to perform another transformation on the coordinates of the descending branch as follows: It can be handled in exactly the same way as the rising branch.
ν _k = 1−u _k ⁻ (formula 12.20)
In these variables, both branches (τ _i ⁻ , ν _i ) and (τ _j ⁺ , u _j ⁺ ) increase monotonically, both have a non-zero slope and behave similarly (convex) Start from the same point (τ = 0, u = 1).

ここで、関数φはＫに直線的に依存し、下記のように定義される。
φ（α，β，Ｋ，τ）≡１＋Ｋζ（α，β，ｔ） （式１２．２２）
ここで、
ζ（α，β，τ）＝α／β（１＋τ／α）β β＞０の場合 （式１２．２３）
＝ｓ≡αｌｎ（１＋τ／α） β＝０の場合
＝α／（１＋β）［（１＋ｓ／α）¹⁺β−１］ −１≦β＜０の場合
関数ζ（α，β，τ）のファミリーを、１５のパラメータの値βに対して図１９に示す。パラメータαは、平面（τ／α，ε／α）上にこの関数をプロットすることによって完全に外される。関数ζ（α，β，τ）は全ての３つの変数の中で連続であり、固定されたαおよびβにおいてτの関数として、τ＝０，ζ’（α，β，０）＝１において単位勾配（unit slope）を有する。このため、τ→０において、関数ζは、αおよびβから独立した極めて単純な動作ζ〜τを示す（そのような動作の過程の領域の尺度は、αおよびβに依存する）。関数ζは次のような重要な特徴を有する、すなわち、βがポイントβ＝０を通過するとき、τ→∞におけるその漸近動作は、β＞０におけるパワー関数ζ〜τβから、β＝０におけるζ〜ｌｎ（τ）を通り、−１＜β＜０の場合の対数のパワー関数ζ〜ｌｎ¹⁺β（τ）まで連続的に変化する。β→−１の場合、ζの動作はもう一度変化し、ζ〜ｌｎ（ｌｎ（τ））になる。β＜１の場合、このファミリーの任意の関数の凸性は同じである。 Since both branches are processed in exactly the same way, we simplify the notation, temporarily omit the superscript ±, and also pair (τ _i ⁻ , ν _i ) or (τ _j ⁺ , Some of u _j ⁺ ) are written as (τ _k , u _k ). The inventors fit the data set {u _k } with {φ (α, β, τ _k )}, whereby {u _k } is expressed as follows.

Here, the function φ depends linearly on K and is defined as follows.
φ (α, β, K, τ) ≡1 + Kζ (α, β, t) (Formula 12.22)
here,
When ζ (α, β, τ) = α / β (1 + τ / α) β β> 0 (Formula 12.23)
= S≡αln (1 + τ / α) When β = 0
= Α / (1 + β) [(1 + s / α) 1+ β-1] -1 ≦ β <0 when the function ζ (α, β, τ) family of FIG relative value beta of 15 parameters 19 Shown in The parameter α is completely removed by plotting this function on the plane (τ / α, ε / α). The function ζ (α, β, τ) is continuous among all three variables, and as a function of τ at fixed α and β, at τ = 0, ζ ′ (α, β, 0) = 1 It has a unit slope. Thus, at τ → 0, the function ζ exhibits a very simple motion ζ˜τ that is independent of α and β (the scale of the region of the process of such motion depends on α and β). The function ζ has the following important features: when β passes through the point β = 0, its asymptotic behavior at τ → ∞ is from the power functions ζ˜τβ at β> 0, and at β = 0 It passes through ζ to ln (τ) and continuously varies from the logarithmic power function ζ to ln ¹⁺ β (τ) when −1 <β <0. In the case of β → −1, the operation of ζ changes once more and becomes ζ˜ln (ln (τ)). If β <1, the convexity of any function in this family is the same.

また、その最小値についての要求事項は、Ｋに対して下記の式を直接発生する。

これらの計算において、発明者らは、ｕ＝１とｕ＝１＋０．４（ｕ_max−１）との間のｕの値を有する全ての隣接するポイントを含むようなＫ１値を用いる。発明者らはこのように、フィッティング手順におけるフィッティングパラメータの数を、２つの連続パラメータαおよびβと１つの離散パラメータγとに減少させた。このため、フィッティング関数は下記のようになる。

パラメータα、βおよびＫ（およびγ）の値は、この場合、下記のｙ空間におけるフィッティングエラーが最小になる条件によって直接決定される。

式（１２．２７）の合計は、１および２に等しいγに対するグリッド（α，β）値上で数値的に評価される。次に、εγ^(Y)を最小値にするα、βおよびγの値が、数値試験により発見される。次に、発見された最小値の近傍のより細かいグリッド上で、計算を繰り返すことができる。
上昇ブランチに対するパラメータα_min、β_minおよびγ_minを求めると、発明者らは稠密なｔグリッド｛ｔ_s，ｓ＝１，２，．．，Ｎ｝を生成し、下記のような間隔の持続時間の対応する値を計算する。

下降ブランチの同様の稠密な表示は、正確に同じ方法で計算することができる。結果として生ずる鳥状の曲線を、図２０のパネルＡおよびパネルＣに示す。実際の絶対エラーおよび相対エラーは、見出しの中で示される。左手側のパネルＢおよびパネルＤは、それぞれがパネルＡおよびパネルＣで示した曲線から結果として生じた、２つの異なる表示のヒステリシス曲線を示す。次に、ヒステリシスループおよびその大きさのその後の計算が、式（１０．１７）によって実施例１１の中で説明されるように実行される。 The inventors require that for a given α and β pair, α and β using equations (Equation 12.22) and equation (Equation 12.23), the parameter K is the point (τ = 0, u = 1 ) Expressly asserts that optimal data fitting is guaranteed within a certain nearby u-space. Let K1 be the number of points that require fitting. The corresponding secondary error is then a function of K, α and β that can be described as follows:

Also, the requirement for the minimum value directly generates the following equation for K:

In these calculations, we use a K1 value that includes all adjacent points with a value of u between u = 1 and u = 1 + 0.4 (u _max −1). The inventors thus reduced the number of fitting parameters in the fitting procedure to two continuous parameters α and β and one discrete parameter γ. For this reason, the fitting function is as follows.

The values of the parameters α, β and K (and γ) are directly determined in this case by the following conditions that minimize the fitting error in the y space.

The sum of equation (12.27) is evaluated numerically on the grid (α, β) values for γ equal to 1 and 2. Next, the values of α, β, and γ that minimize εγ ^(Y) are found by numerical tests. The calculation can then be repeated on a finer grid in the vicinity of the found minimum.
Having determined the parameters α _min , β _min, and γ _min for the rising branch, we have a dense t-grid {t _s , s = 1,2,. . , N} and calculate the corresponding value of the duration of the interval as follows:

A similar dense representation of the descending branch can be calculated in exactly the same way. The resulting bird-like curve is shown in panels A and C of FIG. Actual absolute and relative errors are indicated in the heading. Panel B and Panel D on the left hand side show two different display hysteresis curves resulting from the curves shown in Panel A and Panel C, respectively. Next, the hysteresis loop and subsequent calculation of its magnitude is performed as described in Example 11 by equation (10.17).

[Example 14]
<Creation of RR hysteresis loop by procedures of Examples 10-12>
In addition to the procedure of creating a hysteresis loop on a plane (QT interval vs. RR interval) or equivalent plane, a separate hysteresis of the duration of work of RR interval vs. motion can be introduced and evaluated. The work of this exercise gradually changes in a reciprocating manner during the steps of raising and lowering the exercise. The RR hysteresis can be displayed as a loop on different planes based solely on the analysis of one RR data set {t _RR ⁱ , T _RR ⁱ }. For example, such a loop can be displayed on the (τ ⁱ , T _RR ⁱ ) plane. Here, τ ⁱ = | t _RR ⁱ −t _min |, where t _min is the time corresponding to the peak of the exercise load or the center of the maximum load period. This can be determined based on the numerical techniques described in Examples 10-12. The RR loop can also be introduced on the (W (t _RR ⁱ ), T _RR ⁱ ) or (τ ⁱ , (T _RR ⁱ ) ⁻¹ ) plane. Here, W (t _RR ⁱ ) is a work amount that changes with respect to the exercise stage, that is, time and (T _RR ⁱ ) ⁻¹ indicate the heart rate.

In order to apply the numerical technique from Example 11 (or any other of Example 11 or 12), the same computational steps described in these examples will be essentially repeated. However, both ^{_{QT- {t QT i, T QT}} i} and ^{_{RR- {t RR i, T RR}} i} instead of considering the interval data sets, only one RR data sets, to create a hysteresis loop Are processed numerically over the entire sequence of steps described. In this case, the variable τ ⁱ , (T _RR ⁱ ) ⁻¹ or W (t _RR ⁱ ), together with the first T _RR ⁱ variable, plays the role of a second separate component forming the RR hysteresis plane.

  The foregoing examples are illustrative of the invention and are not to be construed as limiting. The present invention is defined by the following claims, and equivalents of the claims are included therein.

に等しい（実施例１０のセクション７を参照のこと）。
順次移動平均法によるデーター処理のブロック図である（実施例１２、ステップ１〜３）。 実施例１２のステップ１からステップ２にわたる処理結果を示す図である。上パネルは、ステップ１後の処理されたＱＴおよびＲＲのデータセットおよびＱＴ／ＲＲのヒステリシスループを示す（それぞれ左から右へ）。下パネルは、第２の移動平均処理および最終ステップ３の後の同じ平滑化の依存性を示す。 最適化された非線形変換方法における主なステップの、一般的なデーターのフローチャートである。左手側および右手側のボックスは、それぞれＲＲおよびＱＴ間隔に対する同様の処理段階を説明する。 図１５の段階（１ａ／ｂ）〜（３ａ／ｂ）間の１つのデーターサブセット｛ｔ^k，Ｔ^k _RR｝または｛ｔ^k，Ｔ^k _QT｝に対する詳細なデーターのフローチャートを示す図である。先行段階のボックス１〜７は、従来のデーター処理方法の組み合わせを用い、移動平均（１）、最小領域の決定（２）、二次放物線のこの領域のデーターへのフィッティング（３）、結果の整合性のチェック（４）、最小値における最小データーおよび中心データーの検出（５）および（６）、データーの条件付きソーティング（７）を含む。段階（８）〜（１１）は、非線形回帰に対するデュアル非線形変換方法に基づいている。 フィルタ処理したＲＲ間隔のデーターセットの非線形変換を示す図である。パネルＡは本来の（ｔ，ｙ）面上のデーターを示し、最小値は円内のアスタリスクでマークされている。パネルＢおよびＣは、それぞれｊ＝１およびｊ＝２に対する（ｔ，ｕ）面上の変換されたセットを示す。最小値のイメージも、囲ったアスタリスクでマークされている。変換されたデーターセットは、中央部にはっきりした直線部分がある、単調に増加する（平均）曲線の周りに集中していることに注意されたい。 ＱＴ間隔データセットであることを除くと、図１７と同様の図である。 β＝-0.9（下側の曲線）からβ＝0（中間の太い曲線）を通りβ＝0.5（上側の曲線）まで△β＝0.1のステップで変化する、１５の値のパラメータβに対する関数ζ（α，β，τ）のファミリーの適切な縮尺の図である。この関数ζ（α，β，τ）は３つ全ての変数において連続しており、τの関数として、τ＝0，ζ’（α，β，0）＝1において単位勾配を有する。 一人の患者に対するＲＲおよびＱＴデータセットの完全な処理の実施例を示す図である。パネルＡおよびＣは、ＲＲおよびＱＴデータセットおよびそのフィット性を示す。パネルＢは、対応する上昇および下降する曲線並びに（ｆ_RR，Ｔ_QT）面上の閉鎖線を示す。この面上では、そのようなヒステリシスループの区域は、二乗された時間の次元を有する。パネルＤは、（ｆ_RR，Ｔ_QT）面上のヒステリシスループを示す。ここで、ｆ_RR＝１／Ｔ_RRは心拍数であり、ループ区域は無次元数である。パネルＡに対する全エラーは2.2％であり、パネルＣについては0.8％である。 最大の仕事量が２８Ｗ（ワット）の急激な運動停止前の4.5分の仕事量を行った６０才の冠動脈疾患「ＣＡＤ」の患者についてのＲＲ間隔データセットである。 図２１と同じＣＡＤ患者の、仕事量が4.5分の急激な運動停止後の最初の1.5分の回復期間に対するＲＲ間隔データセットである。ＨＲの回復が、９拍／分という遅い準定常的な割合で発生している。 １３０Ｗの最大仕事量の後の２分の低い２０Ｗの回復仕事量の後の運動をほとんど急に停止する前の１４分の仕事量を行った後の、５０才の被験者に対するＲＲ間隔データセットである。 図２３と同じ５０才の被験者の、１３０Ｗの最大仕事量の後の２分の低い２０Ｗの回復仕事量の後の運動をほぼ急に停止した後の、最初の１．５分の回復期間に対するＲＲ間隔のデーターセットである。 冠動脈疾患の患者および冠動脈疾患ではない患者に対するＲＲ間隔およびＱＴ間隔データセット並びに計算されたヒステリシスループである。 冠動脈疾患の患者および冠動脈疾患ではない患者に対するＲＲ間隔およびＱＴ間隔のデーターセット並びに計算されたヒステリシスループである。 FIG. 2 is a schematic diagram of action potentials summed over that volume in the myocardium and the generated electrocardiogram (ECG) recorded on the human body surface. FIG. 4 shows mathematical formulas used in a simplified mathematical model of periodic excitement. It is a figure which shows a periodic excitement wave. (The computer generated action potential u and instantaneous threshold v are shown using a simplified mathematical model, which are described in FIG. 2A.) FIG. 6 shows a family of four synthetic dispersion-recovery curves corresponding to four values of the media excitability threshold. Fig. 2 is a block diagram of an apparatus for performing the method according to the present invention. FIG. 3 is a block diagram of processing steps for data collection and analysis of the present invention. FIG. 4 is another block diagram of processing steps for data collection and analysis of the present invention. FIG. 6 shows a hysteresis loop of QT interval versus RR interval obtained from experiments on two healthy male subjects (23 years old, thick line and 47 years old, thin line) plotted on the synthetic dispersion-recovery curve surface. is there. It is a figure which shows the example of the hysteresis of a QT-RR space | interval of two male test subjects. One has a drop in the ST segment of the conventional ECG (thin line) and the other has a history of myocardial infarction 12 years before the trial (thick line). The generation of these curves is described in more detail in the following specification. It is a figure which shows the hysteresis loop of the QT-RR space | interval with respect to a male test subject while explaining the sensitivity of this invention. The first loop (thick line) corresponds to the initial test while a ST segment drop was observed on a conventional ECG, and the second loop, shown as a thin line, was measured after a regular exercise cycle. It was done. FIG. 6 illustrates a comparative analysis of cardiac ischemia based on a specific example of a normalized measure of the area of the hysteresis loop. _{<CII> = (CII -CII min} ) / (CII max - CII min) ( "CII" means "heart ischemia index"). O ₁ , X ₁ and Y ₁ represent subject data. X _i represents data (0.28 to 0.35) collected from one subject in a series of tests (day / night test, running / walking, test interval is about 2 months), and peak exercise heart rate is 120 To 135. Y _i represents data (0.46 to 0.86) collected from one subject in a series of tests (running / walking, interval between tests before and after the period of regular exercise phase is 6 weeks), and peak of exercise The heart rate ranges from 122 to 146. Black bars indicate areas where the conventional ST descent method does not detect cardiac ischemia (<CII> is less than 0.70). Conventional methods, a high white bars _{(Y 2, Y 3, O} 7: <CII> 0.70 greater) can detect cardiac ischemia only in a very narrow range indicated by. FIG. 7 shows a general peripheral nervous system and hormonal rapid control adjustment to QT and RR intervals for a sudden cessation of movement (ie, a sudden onset of the resting phase). FIG. 4 shows a general slow (quasi-stationary) QT and RR interval adjustment measured while gradually increasing and gradually decreasing cardiac stimulation. It is a figure which shows the block diagram of the data processing by the method of optimizing the connection of a moving average, an exponential function, and a fitting of a polynomial (Example 11, steps 1-8). It is a figure which shows the result of the process over the whole steps 1-8 of Example 11. FIG. The upper panel shows the QT and RR data sets processed from steps 1 to 3 (respectively from left to right) and the QT / RR hysteresis loop after step 1. The exponential fitting curve (step 3) is shown in gray in the first two panels. The lower panel shows the same smoothing dependency after processing from step 4 to final step 8. Where CII (see lower right panel) is the ratio

(See Section 7 of Example 10).
It is a block diagram of the data processing by a sequential moving average method (Example 12, steps 1-3). It is a figure which shows the process result from step 1 of Example 12 to step 2. FIG. The upper panel shows the processed QT and RR data sets and QT / RR hysteresis loop after step 1 (from left to right, respectively). The lower panel shows the same smoothing dependency after the second moving average process and final step 3. 2 is a general data flow chart of the main steps in an optimized nonlinear transformation method. The left hand and right hand boxes describe similar processing steps for the RR and QT intervals, respectively. FIG. 16 shows a detailed data flow chart for one data subset {t ^k , T ^k _RR } or {t ^k , T ^k _QT } between the steps (1a / b) to (3a / b) of FIG. 15. The preceding boxes 1-7 use a combination of conventional data processing methods, moving average (1), determination of minimum area (2), fitting of secondary parabola to data in this area (3), Consistency check (4), detection of minimum and center data at minimum value (5) and (6), conditional sorting of data (7). Steps (8)-(11) are based on a dual nonlinear transformation method for nonlinear regression. It is a figure which shows the nonlinear transformation of the data set of the filtered RR interval. Panel A shows the data on the original (t, y) plane, and the minimum value is marked with an asterisk in the circle. Panels B and C show the transformed set on the (t, u) plane for j = 1 and j = 2, respectively. The minimum image is also marked with an enclosed asterisk. Note that the transformed data set is centered around a monotonically increasing (average) curve with a sharp straight line in the middle. FIG. 18 is a view similar to FIG. 17 except that it is a QT interval data set. A function ζ for a parameter β of 15 values that changes in steps of Δβ = 0.1 from β = −0.9 (lower curve) through β = 0 (middle thick curve) to β = 0.5 (upper curve) FIG. 4 is a diagram of an appropriate scale of a family of (α, β, τ). This function ζ (α, β, τ) is continuous in all three variables, and has a unit gradient at τ = 0, ζ ′ (α, β, 0) = 1 as a function of τ. FIG. 4 shows an example of complete processing of RR and QT data sets for a single patient. Panels A and C show the RR and QT data sets and their fit. Panel B shows the corresponding rising and falling curves and the closing line on the (f _RR , T _QT ) plane. On this plane, the area of such a hysteresis loop has a squared time dimension. Panel D shows a hysteresis loop on the (f _RR , T _QT ) plane. Here, f _RR = 1 / T _RR is the heart rate, and the loop area is a dimensionless number. The total error for panel A is 2.2% and for panel C is 0.8%. RR interval data set for a 60-year-old coronary artery disease “CAD” patient who performed 4.5 minutes of work prior to a sudden cessation of exercise with a maximum work of 28 W (Watt). FIG. 22 is an RR interval data set for the same CAD patient as in FIG. 21 for the first 1.5-minute recovery period after a sudden cessation of work for 4.5 minutes. HR recovery occurs at a slow quasi-stationary rate of 9 beats / minute. In the RR interval data set for a 50 year old subject after 14 minutes of work before ending the exercise after 2 minutes low 20W recovery work after a maximum work of 130W almost abruptly is there. For the same 1.5-year recovery period of the same 50-year-old subject as in FIG. 23, after almost abruptly stopping the exercise after 2W low 20W recovery work after 130W maximum work RR interval data set. RR and QT interval data sets and calculated hysteresis loops for patients with and without coronary artery disease. RR and QT interval data sets and calculated hysteresis loops for patients with and without coronary artery disease.

Claims (46)

(A) collecting a first RR interval data set from the subject during a phase in which the heart rate gradually increases to a predetermined threshold of at least 130 beats / minute;
(B) collecting a second RR interval data set from the subject during a phase in which the heart rate gradually decreases;
(C) comparing the first RR interval data set and the second RR interval data set to determine a difference between the data sets;
(D) From the comparison in step (c), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the health condition of the heart or cardiovascular is decreased Generating a measure of cardiac ischemia while the subject is being stimulated to provide a measure of the heart or cardiovascular health status of the subject, A method for assessing ischemia. The predetermined heart rate threshold is:
Formula PT = (220 ± V bpm) -age
The PT according to claim 1, wherein PT is a predetermined heart rate threshold, V is 20, bpm is beats per minute, and age is the age of the subject. the method of. (E) collecting the first and second QT interval data sets from the subject simultaneously with steps (a) and (b);
(F) comparing the first QT interval data set with the second QT interval data set to determine a difference between the first and second QT interval data sets;
(G) From the comparison step (f), a large difference between the first and second data sets indicates that the subject's cardiac ischemia is large and the heart or cardiovascular health condition is reduced. Generating a measure of cardiac ischemia while the subject is being stimulated, the method of claim 1, further comprising:   The method of claim 1, wherein the step of gradually decreasing the heart rate is performed after a cool-down step of reducing exercise load.   5. The method of claim 4, wherein the step of reducing exercise load is about 2 minutes in duration.   2. The method of claim 1, wherein the subject does not suffer from coronary artery disease.   The method of claim 1, wherein the subject suffers from coronary artery disease.   2. The method of claim 1, wherein the subject suffers from moderate coronary artery disease.   The method of claim 1, wherein the first and second RR interval data sets are collected under quasi-steady state.   The method of claim 2, wherein the first and second RR interval data sets are collected under quasi-steady state.   2. The method of claim 1, wherein each of the step of gradually increasing the heart rate and the step of gradually decreasing the heart rate each have a duration of at least 3 minutes.   The method of claim 1, wherein the step of gradually increasing the heart rate and the step of gradually decreasing the heart rate are performed together for a total time of 6 to 40 minutes.   Both the stage in which the heart rate gradually increases and the stage in which the heart rate gradually decreases are performed between the peak heart rate and the minimum heart rate, and the heart rate and the stage in which the heart rate gradually increases The method of claim 1, wherein the peak heart rate is the same for both the gradually decreasing steps.   14. The method of claim 13, wherein the minimum heart rate for both the step of gradually increasing heart rate and the step of gradually decreasing heart rate is substantially the same.   The method of claim 1, wherein the step of gradually decreasing heart rate is performed at at least three different heart rate stimulation levels.   The method of claim 15, wherein the step of gradually increasing heart rate is performed at at least three different heart rate stimulation levels.   The method according to claim 1, wherein the step of gradually increasing the heart rate and the step of gradually decreasing the heart rate are performed continuously in time.   The method of claim 1, wherein the step of gradually increasing the heart rate and the step of gradually decreasing the heart rate are performed at an interval in time.   The method of claim 1, wherein the generating step is performed by generating a curve from the respective data set.   20. The method of claim 19, wherein the generating step is performed by comparing the shape of the curve from a data set.   The method of claim 19, wherein the generating step is performed by determining a size of a region between the curves. 20. The method of claim 19, wherein the generating step is performed by both comparing the shape of the curves from a data set and determining the size of the area between the curves. The method of claim 19, further comprising displaying the curve.   The method of claim 1, wherein the heart rate exceeds 150 beats / minute during the phase of gradually increasing heart rate. (E) comparing the measure of cardiac ischemia during application of the stimulus with at least one reference value;
2. The method of claim 1, further comprising: (f) generating a quantitative indication of heart or cardiovascular health status for the subject from the comparison of step (e). (G) treating the subject with cardiovascular therapy;
(H) Repeat steps (a) to (f) to evaluate the effect of the cardiovascular therapy and reduce differences between the data sets from before the therapy to after the therapy 26. The method of claim 25, further comprising: repeating the step of indicating that the cardiovascular therapy is improving the heart health of the subject.   The cardiovascular therapy is selected from the group consisting of aerobic exercises, strength building, diet changes, nutritional supplements, weight loss, stress relief, smoking cessation, drug treatment, surgical treatment, and combinations thereof 27. The method of claim 26.   26. The method of claim 25, further comprising assessing, from the quantitative display, a risk of the subject facing a cardiac accident related to a future ischemia. The computer readable program code is
(A) a computer readable program code that collects a first RR interval data set from the subject during a phase in which the heart rate gradually increases to a predetermined threshold of at least 130 beats / minute;
(B) a computer readable program code that collects a second RR interval data set from the subject during a phase in which the heart rate gradually decreases;
(C) a computer readable program code for comparing the first RR interval data set and the second RR interval data set to determine a difference between the data sets;
(D) From the comparison in the computer readable program code (c), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the heart or cardiovascular A computer readable program code for generating a measure of cardiac ischemia while the subject is being stimulated, indicating a low health state, and the subject's heart or cardiovascular health state A computer program product comprising a computer usable recording medium having a computer readable program code embedded in the medium for providing a measure of the above and assessing cardiac ischemia in said subject. (E) step (a) and step (b) simultaneously, computer readable program code for collecting first and second RR interval data sets from said subject;
(F) a computer readable program code for comparing the first QT interval data set and the second QT interval data set to determine a difference between the data sets;
(G) From the comparison in the computer readable program code (f), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the heart or cardiovascular 30. The computer program product of claim 29, further comprising: computer readable program code for generating a measure of cardiac ischemia while the subject is being stimulated indicating a low health condition. .   30. The computer program product of claim 29, wherein the generated program code (g) readable by the computer includes computer readable program code for generating a curve from each of the data sets.   The computer program product of claim 31, wherein the generated program code (g) readable by the computer comprises computer readable program code for comparing the shape of the curve from a data set.   32. The computer program according to claim 31, wherein the generated program code readable by the computer comprises a computer readable program code for determining the size of the area between the curves. Product.   The generated computer readable program code (g) comprises computer readable program code for comparing the shape of the curves from a data set to determine the size of the area between the curves. 32. The computer program product of claim 31.   32. The computer program product of claim 31, further comprising computer readable program code for displaying the curve. (E) a computer readable program code for comparing the measure of cardiac ischemia during application of the stimulus with at least one reference value;
30. A computer readable program code for generating, from the comparison of (e), a quantitative display of heart or cardiovascular health for the subject. The computer program product described in 1.   37. The computer program further comprising: computer readable program code for assessing, from the quantitative display, a risk of the subject experiencing a cardiac event associated with future ischemia. The computer program product described in 1. (A) means for collecting a first RR interval data set from the subject during a phase in which the heart rate gradually increases to a predetermined threshold of at least 130 beats / minute;
(B) means for collecting a second RR interval data set from the subject during a phase in which the heart rate gradually decreases;
(C) means for comparing the first RR interval data set and the second RR interval data set to determine a difference between the data sets;
(D) From the comparison of the computer readable program code (c), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the heart or cardiovascular Means for generating a measure of cardiac ischemia while the subject is being stimulated, indicating that the subject's health condition is reduced, and comparing the subject's cardiac ischemia, Providing a measure of the heart or cardiovascular health of the subject and assessing cardiac ischemia in the subject. (E) means for collecting said first and second QT interval data sets from said subject simultaneously with said steps (a) and (b);
(F) means for comparing the first QT interval data set and the second QT interval data set to determine a difference between the first and second QT interval data sets;
(G) From the comparison (f), if the difference between the first and second data sets is large, the subject's cardiac ischemia is large and the heart or cardiovascular health condition is reduced. 39. The system of claim 38, further comprising: means for generating a measure of cardiac ischemia while the subject is being stimulated.   39. The system of claim 38, wherein the generating means (g) includes means for generating a curve from each of the data sets.   41. The system of claim 40, wherein the generating means (g) comprises means for comparing the shape of the curve from a data set.   41. The system of claim 40, wherein the generating means (g) includes means for determining a size of a region between the curves.   41. The system of claim 40, wherein the generating means (g) includes means for comparing the shape of the curves from a data set to determine the size of the area between the curves.   41. The system of claim 40, further comprising means for displaying the curve. (E) means for comparing said measure of cardiac ischemia during application of the stimulus with at least one reference value;
And (f) means for generating a quantitative indication of heart or cardiovascular health status for the subject from the comparison of (e).   46. The system of claim 45, further comprising means for assessing, from the quantitative display, a risk of the subject facing a cardiac accident related to future ischemia.

Images